fs/ext4/ioctl.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * linux/fs/ext4/ioctl.c
   4  *
   5  * Copyright (C) 1993, 1994, 1995
   6  * Remy Card ([email protected])
   7  * Laboratoire MASI - Institut Blaise Pascal
   8  * Universite Pierre et Marie Curie (Paris VI)
   9  */
  10
  11 #include <linux/fs.h>
  12 #include <linux/capability.h>
  13 #include <linux/time.h>
  14 #include <linux/compat.h>
  15 #include <linux/mount.h>
  16 #include <linux/file.h>
  17 #include <linux/quotaops.h>
  18 #include <linux/random.h>
  19 #include <linux/uaccess.h>
  20 #include <linux/delay.h>
  21 #include <linux/iversion.h>
  22 #include <linux/fileattr.h>
  23 #include <linux/uuid.h>
  24 #include "ext4_jbd2.h"
  25 #include "ext4.h"
  26 #include <linux/fsmap.h>
  27 #include "fsmap.h"
  28 #include <trace/events/ext4.h>
  29
  30 typedef void ext4_update_sb_callback(struct ext4_super_block *es,
  31                                        const void *arg);
  32
  33 /*
  34  * Superblock modification callback function for changing file system
  35  * label
  36  */
  37 static void ext4_sb_setlabel(struct ext4_super_block *es, const void *arg)
  38 {
  39         /* Sanity check, this should never happen */
  40         BUILD_BUG_ON(sizeof(es->s_volume_name) < EXT4_LABEL_MAX);
  41
  42         memcpy(es->s_volume_name, (char *)arg, EXT4_LABEL_MAX);
  43 }
  44
  45 /*
  46  * Superblock modification callback function for changing file system
  47  * UUID.
  48  */
  49 static void ext4_sb_setuuid(struct ext4_super_block *es, const void *arg)
  50 {
  51         memcpy(es->s_uuid, (__u8 *)arg, UUID_SIZE);
  52 }
  53
  54 static
  55 int ext4_update_primary_sb(struct super_block *sb, handle_t *handle,
  56                            ext4_update_sb_callback func,
  57                            const void *arg)
  58 {
  59         int err = 0;
  60         struct ext4_sb_info *sbi = EXT4_SB(sb);
  61         struct buffer_head *bh = sbi->s_sbh;
  62         struct ext4_super_block *es = sbi->s_es;
  63
  64         trace_ext4_update_sb(sb, bh->b_blocknr, 1);
  65
  66         BUFFER_TRACE(bh, "get_write_access");
  67         err = ext4_journal_get_write_access(handle, sb,
  68                                             bh,
  69                                             EXT4_JTR_NONE);
  70         if (err)
  71                 goto out_err;
  72
  73         lock_buffer(bh);
  74         func(es, arg);
  75         ext4_superblock_csum_set(sb);
  76         unlock_buffer(bh);
  77
  78         if (buffer_write_io_error(bh) || !buffer_uptodate(bh)) {
  79                 ext4_msg(sbi->s_sb, KERN_ERR, "previous I/O error to "
  80                          "superblock detected");
  81                 clear_buffer_write_io_error(bh);
  82                 set_buffer_uptodate(bh);
  83         }
  84
  85         err = ext4_handle_dirty_metadata(handle, NULL, bh);
  86         if (err)
  87                 goto out_err;
  88         err = sync_dirty_buffer(bh);
  89 out_err:
  90         ext4_std_error(sb, err);
  91         return err;
  92 }
  93
  94 /*
  95  * Update one backup superblock in the group 'grp' using the callback
  96  * function 'func' and argument 'arg'. If the handle is NULL the
  97  * modification is not journalled.
  98  *
  99  * Returns: 0 when no modification was done (no superblock in the group)
 100  *          1 when the modification was successful
 101  *         <0 on error
 102  */
 103 static int ext4_update_backup_sb(struct super_block *sb,
 104                                  handle_t *handle, ext4_group_t grp,
 105                                  ext4_update_sb_callback func, const void *arg)
 106 {
 107         int err = 0;
 108         ext4_fsblk_t sb_block;
 109         struct buffer_head *bh;
 110         unsigned long offset = 0;
 111         struct ext4_super_block *es;
 112
 113         if (!ext4_bg_has_super(sb, grp))
 114                 return 0;
 115
 116         /*
 117          * For the group 0 there is always 1k padding, so we have
 118          * either adjust offset, or sb_block depending on blocksize
 119          */
 120         if (grp == 0) {
 121                 sb_block = 1 * EXT4_MIN_BLOCK_SIZE;
 122                 offset = do_div(sb_block, sb->s_blocksize);
 123         } else {
 124                 sb_block = ext4_group_first_block_no(sb, grp);
 125                 offset = 0;
 126         }
 127
 128         trace_ext4_update_sb(sb, sb_block, handle ? 1 : 0);
 129
 130         bh = ext4_sb_bread(sb, sb_block, 0);
 131         if (IS_ERR(bh))
 132                 return PTR_ERR(bh);
 133
 134         if (handle) {
 135                 BUFFER_TRACE(bh, "get_write_access");
 136                 err = ext4_journal_get_write_access(handle, sb,
 137                                                     bh,
 138                                                     EXT4_JTR_NONE);
 139                 if (err)
 140                         goto out_bh;
 141         }
 142
 143         es = (struct ext4_super_block *) (bh->b_data + offset);
 144         lock_buffer(bh);
 145         if (ext4_has_metadata_csum(sb) &&
 146             es->s_checksum != ext4_superblock_csum(sb, es)) {
 147                 ext4_msg(sb, KERN_ERR, "Invalid checksum for backup "
 148                 "superblock %llu", sb_block);
 149                 unlock_buffer(bh);
 150                 goto out_bh;
 151         }
 152         func(es, arg);
 153         if (ext4_has_metadata_csum(sb))
 154                 es->s_checksum = ext4_superblock_csum(sb, es);
 155         set_buffer_uptodate(bh);
 156         unlock_buffer(bh);
 157
 158         if (handle) {
 159                 err = ext4_handle_dirty_metadata(handle, NULL, bh);
 160                 if (err)
 161                         goto out_bh;
 162         } else {
 163                 BUFFER_TRACE(bh, "marking dirty");
 164                 mark_buffer_dirty(bh);
 165         }
 166         err = sync_dirty_buffer(bh);
 167
 168 out_bh:
 169         brelse(bh);
 170         ext4_std_error(sb, err);
 171         return (err) ? err : 1;
 172 }
 173
 174 /*
 175  * Update primary and backup superblocks using the provided function
 176  * func and argument arg.
 177  *
 178  * Only the primary superblock and at most two backup superblock
 179  * modifications are journalled; the rest is modified without journal.
 180  * This is safe because e2fsck will re-write them if there is a problem,
 181  * and we're very unlikely to ever need more than two backups.
 182  */
 183 static
 184 int ext4_update_superblocks_fn(struct super_block *sb,
 185                                ext4_update_sb_callback func,
 186                                const void *arg)
 187 {
 188         handle_t *handle;
 189         ext4_group_t ngroups;
 190         unsigned int three = 1;
 191         unsigned int five = 5;
 192         unsigned int seven = 7;
 193         int err = 0, ret, i;
 194         ext4_group_t grp, primary_grp;
 195         struct ext4_sb_info *sbi = EXT4_SB(sb);
 196
 197         /*
 198          * We can't update superblocks while the online resize is running
 199          */
 200         if (test_and_set_bit_lock(EXT4_FLAGS_RESIZING,
 201                                   &sbi->s_ext4_flags)) {
 202                 ext4_msg(sb, KERN_ERR, "Can't modify superblock while"
 203                          "performing online resize");
 204                 return -EBUSY;
 205         }
 206
 207         /*
 208          * We're only going to update primary superblock and two
 209          * backup superblocks in this transaction.
 210          */
 211         handle = ext4_journal_start_sb(sb, EXT4_HT_MISC, 3);
 212         if (IS_ERR(handle)) {
 213                 err = PTR_ERR(handle);
 214                 goto out;
 215         }
 216
 217         /* Update primary superblock */
 218         err = ext4_update_primary_sb(sb, handle, func, arg);
 219         if (err) {
 220                 ext4_msg(sb, KERN_ERR, "Failed to update primary "
 221                          "superblock");
 222                 goto out_journal;
 223         }
 224
 225         primary_grp = ext4_get_group_number(sb, sbi->s_sbh->b_blocknr);
 226         ngroups = ext4_get_groups_count(sb);
 227
 228         /*
 229          * Update backup superblocks. We have to start from group 0
 230          * because it might not be where the primary superblock is
 231          * if the fs is mounted with -o sb=<backup_sb_block>
 232          */
 233         i = 0;
 234         grp = 0;
 235         while (grp < ngroups) {
 236                 /* Skip primary superblock */
 237                 if (grp == primary_grp)
 238                         goto next_grp;
 239
 240                 ret = ext4_update_backup_sb(sb, handle, grp, func, arg);
 241                 if (ret < 0) {
 242                         /* Ignore bad checksum; try to update next sb */
 243                         if (ret == -EFSBADCRC)
 244                                 goto next_grp;
 245                         err = ret;
 246                         goto out_journal;
 247                 }
 248
 249                 i += ret;
 250                 if (handle && i > 1) {
 251                         /*
 252                          * We're only journalling primary superblock and
 253                          * two backup superblocks; the rest is not
 254                          * journalled.
 255                          */
 256                         err = ext4_journal_stop(handle);
 257                         if (err)
 258                                 goto out;
 259                         handle = NULL;
 260                 }
 261 next_grp:
 262                 grp = ext4_list_backups(sb, &three, &five, &seven);
 263         }
 264
 265 out_journal:
 266         if (handle) {
 267                 ret = ext4_journal_stop(handle);
 268                 if (ret && !err)
 269                         err = ret;
 270         }
 271 out:
 272         clear_bit_unlock(EXT4_FLAGS_RESIZING, &sbi->s_ext4_flags);
 273         smp_mb__after_atomic();
 274         return err ? err : 0;
 275 }
 276
 277 /*
 278  * Swap memory between @a and @b for @len bytes.
 279  *
 280  * @a:          pointer to first memory area
 281  * @b:          pointer to second memory area
 282  * @len:        number of bytes to swap
 283  *
 284  */
 285 static void memswap(void *a, void *b, size_t len)
 286 {
 287         unsigned char *ap, *bp;
 288
 289         ap = (unsigned char *)a;
 290         bp = (unsigned char *)b;
 291         while (len-- > 0) {
 292                 swap(*ap, *bp);
 293                 ap++;
 294                 bp++;
 295         }
 296 }
 297
 298 /*
 299  * Swap i_data and associated attributes between @inode1 and @inode2.
 300  * This function is used for the primary swap between inode1 and inode2
 301  * and also to revert this primary swap in case of errors.
 302  *
 303  * Therefore you have to make sure, that calling this method twice
 304  * will revert all changes.
 305  *
 306  * @inode1:     pointer to first inode
 307  * @inode2:     pointer to second inode
 308  */
 309 static void swap_inode_data(struct inode *inode1, struct inode *inode2)
 310 {
 311         loff_t isize;
 312         struct ext4_inode_info *ei1;
 313         struct ext4_inode_info *ei2;
 314         unsigned long tmp;
 315
 316         ei1 = EXT4_I(inode1);
 317         ei2 = EXT4_I(inode2);
 318
 319         swap(inode1->i_version, inode2->i_version);
 320         swap(inode1->i_atime, inode2->i_atime);
 321         swap(inode1->i_mtime, inode2->i_mtime);
 322
 323         memswap(ei1->i_data, ei2->i_data, sizeof(ei1->i_data));
 324         tmp = ei1->i_flags & EXT4_FL_SHOULD_SWAP;
 325         ei1->i_flags = (ei2->i_flags & EXT4_FL_SHOULD_SWAP) |
 326                 (ei1->i_flags & ~EXT4_FL_SHOULD_SWAP);
 327         ei2->i_flags = tmp | (ei2->i_flags & ~EXT4_FL_SHOULD_SWAP);
 328         swap(ei1->i_disksize, ei2->i_disksize);
 329         ext4_es_remove_extent(inode1, 0, EXT_MAX_BLOCKS);
 330         ext4_es_remove_extent(inode2, 0, EXT_MAX_BLOCKS);
 331
 332         isize = i_size_read(inode1);
 333         i_size_write(inode1, i_size_read(inode2));
 334         i_size_write(inode2, isize);
 335 }
 336
 337 void ext4_reset_inode_seed(struct inode *inode)
 338 {
 339         struct ext4_inode_info *ei = EXT4_I(inode);
 340         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
 341         __le32 inum = cpu_to_le32(inode->i_ino);
 342         __le32 gen = cpu_to_le32(inode->i_generation);
 343         __u32 csum;
 344
 345         if (!ext4_has_metadata_csum(inode->i_sb))
 346                 return;
 347
 348         csum = ext4_chksum(sbi, sbi->s_csum_seed, (__u8 *)&inum, sizeof(inum));
 349         ei->i_csum_seed = ext4_chksum(sbi, csum, (__u8 *)&gen, sizeof(gen));
 350 }
 351
 352 /*
 353  * Swap the information from the given @inode and the inode
 354  * EXT4_BOOT_LOADER_INO. It will basically swap i_data and all other
 355  * important fields of the inodes.
 356  *
 357  * @sb:         the super block of the filesystem
 358  * @idmap:      idmap of the mount the inode was found from
 359  * @inode:      the inode to swap with EXT4_BOOT_LOADER_INO
 360  *
 361  */
 362 static long swap_inode_boot_loader(struct super_block *sb,
 363                                 struct mnt_idmap *idmap,
 364                                 struct inode *inode)
 365 {
 366         handle_t *handle;
 367         int err;
 368         struct inode *inode_bl;
 369         struct ext4_inode_info *ei_bl;
 370         qsize_t size, size_bl, diff;
 371         blkcnt_t blocks;
 372         unsigned short bytes;
 373
 374         inode_bl = ext4_iget(sb, EXT4_BOOT_LOADER_INO,
 375                         EXT4_IGET_SPECIAL | EXT4_IGET_BAD);
 376         if (IS_ERR(inode_bl))
 377                 return PTR_ERR(inode_bl);
 378         ei_bl = EXT4_I(inode_bl);
 379
 380         /* Protect orig inodes against a truncate and make sure,
 381          * that only 1 swap_inode_boot_loader is running. */
 382         lock_two_nondirectories(inode, inode_bl);
 383
 384         if (inode->i_nlink != 1 || !S_ISREG(inode->i_mode) ||
 385             IS_SWAPFILE(inode) || IS_ENCRYPTED(inode) ||
 386             (EXT4_I(inode)->i_flags & EXT4_JOURNAL_DATA_FL) ||
 387             ext4_has_inline_data(inode)) {
 388                 err = -EINVAL;
 389                 goto journal_err_out;
 390         }
 391
 392         if (IS_RDONLY(inode) || IS_APPEND(inode) || IS_IMMUTABLE(inode) ||
 393             !inode_owner_or_capable(idmap, inode) ||
 394             !capable(CAP_SYS_ADMIN)) {
 395                 err = -EPERM;
 396                 goto journal_err_out;
 397         }
 398
 399         filemap_invalidate_lock(inode->i_mapping);
 400         err = filemap_write_and_wait(inode->i_mapping);
 401         if (err)
 402                 goto err_out;
 403
 404         err = filemap_write_and_wait(inode_bl->i_mapping);
 405         if (err)
 406                 goto err_out;
 407
 408         /* Wait for all existing dio workers */
 409         inode_dio_wait(inode);
 410         inode_dio_wait(inode_bl);
 411
 412         truncate_inode_pages(&inode->i_data, 0);
 413         truncate_inode_pages(&inode_bl->i_data, 0);
 414
 415         handle = ext4_journal_start(inode_bl, EXT4_HT_MOVE_EXTENTS, 2);
 416         if (IS_ERR(handle)) {
 417                 err = -EINVAL;
 418                 goto err_out;
 419         }
 420         ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_SWAP_BOOT, handle);
 421
 422         /* Protect extent tree against block allocations via delalloc */
 423         ext4_double_down_write_data_sem(inode, inode_bl);
 424
 425         if (is_bad_inode(inode_bl) || !S_ISREG(inode_bl->i_mode)) {
 426                 /* this inode has never been used as a BOOT_LOADER */
 427                 set_nlink(inode_bl, 1);
 428                 i_uid_write(inode_bl, 0);
 429                 i_gid_write(inode_bl, 0);
 430                 inode_bl->i_flags = 0;
 431                 ei_bl->i_flags = 0;
 432                 inode_set_iversion(inode_bl, 1);
 433                 i_size_write(inode_bl, 0);
 434                 EXT4_I(inode_bl)->i_disksize = inode_bl->i_size;
 435                 inode_bl->i_mode = S_IFREG;
 436                 if (ext4_has_feature_extents(sb)) {
 437                         ext4_set_inode_flag(inode_bl, EXT4_INODE_EXTENTS);
 438                         ext4_ext_tree_init(handle, inode_bl);
 439                 } else
 440                         memset(ei_bl->i_data, 0, sizeof(ei_bl->i_data));
 441         }
 442
 443         err = dquot_initialize(inode);
 444         if (err)
 445                 goto err_out1;
 446
 447         size = (qsize_t)(inode->i_blocks) * (1 << 9) + inode->i_bytes;
 448         size_bl = (qsize_t)(inode_bl->i_blocks) * (1 << 9) + inode_bl->i_bytes;
 449         diff = size - size_bl;
 450         swap_inode_data(inode, inode_bl);
 451
 452         inode->i_ctime = inode_bl->i_ctime = current_time(inode);
 453         inode_inc_iversion(inode);
 454
 455         inode->i_generation = get_random_u32();
 456         inode_bl->i_generation = get_random_u32();
 457         ext4_reset_inode_seed(inode);
 458         ext4_reset_inode_seed(inode_bl);
 459
 460         ext4_discard_preallocations(inode, 0);
 461
 462         err = ext4_mark_inode_dirty(handle, inode);
 463         if (err < 0) {
 464                 /* No need to update quota information. */
 465                 ext4_warning(inode->i_sb,
 466                         "couldn't mark inode #%lu dirty (err %d)",
 467                         inode->i_ino, err);
 468                 /* Revert all changes: */
 469                 swap_inode_data(inode, inode_bl);
 470                 ext4_mark_inode_dirty(handle, inode);
 471                 goto err_out1;
 472         }
 473
 474         blocks = inode_bl->i_blocks;
 475         bytes = inode_bl->i_bytes;
 476         inode_bl->i_blocks = inode->i_blocks;
 477         inode_bl->i_bytes = inode->i_bytes;
 478         err = ext4_mark_inode_dirty(handle, inode_bl);
 479         if (err < 0) {
 480                 /* No need to update quota information. */
 481                 ext4_warning(inode_bl->i_sb,
 482                         "couldn't mark inode #%lu dirty (err %d)",
 483                         inode_bl->i_ino, err);
 484                 goto revert;
 485         }
 486
 487         /* Bootloader inode should not be counted into quota information. */
 488         if (diff > 0)
 489                 dquot_free_space(inode, diff);
 490         else
 491                 err = dquot_alloc_space(inode, -1 * diff);
 492
 493         if (err < 0) {
 494 revert:
 495                 /* Revert all changes: */
 496                 inode_bl->i_blocks = blocks;
 497                 inode_bl->i_bytes = bytes;
 498                 swap_inode_data(inode, inode_bl);
 499                 ext4_mark_inode_dirty(handle, inode);
 500                 ext4_mark_inode_dirty(handle, inode_bl);
 501         }
 502
 503 err_out1:
 504         ext4_journal_stop(handle);
 505         ext4_double_up_write_data_sem(inode, inode_bl);
 506
 507 err_out:
 508         filemap_invalidate_unlock(inode->i_mapping);
 509 journal_err_out:
 510         unlock_two_nondirectories(inode, inode_bl);
 511         iput(inode_bl);
 512         return err;
 513 }
 514
 515 /*
 516  * If immutable is set and we are not clearing it, we're not allowed to change
 517  * anything else in the inode.  Don't error out if we're only trying to set
 518  * immutable on an immutable file.
 519  */
 520 static int ext4_ioctl_check_immutable(struct inode *inode, __u32 new_projid,
 521                                       unsigned int flags)
 522 {
 523         struct ext4_inode_info *ei = EXT4_I(inode);
 524         unsigned int oldflags = ei->i_flags;
 525
 526         if (!(oldflags & EXT4_IMMUTABLE_FL) || !(flags & EXT4_IMMUTABLE_FL))
 527                 return 0;
 528
 529         if ((oldflags & ~EXT4_IMMUTABLE_FL) != (flags & ~EXT4_IMMUTABLE_FL))
 530                 return -EPERM;
 531         if (ext4_has_feature_project(inode->i_sb) &&
 532             __kprojid_val(ei->i_projid) != new_projid)
 533                 return -EPERM;
 534
 535         return 0;
 536 }
 537
 538 static void ext4_dax_dontcache(struct inode *inode, unsigned int flags)
 539 {
 540         struct ext4_inode_info *ei = EXT4_I(inode);
 541
 542         if (S_ISDIR(inode->i_mode))
 543                 return;
 544
 545         if (test_opt2(inode->i_sb, DAX_NEVER) ||
 546             test_opt(inode->i_sb, DAX_ALWAYS))
 547                 return;
 548
 549         if ((ei->i_flags ^ flags) & EXT4_DAX_FL)
 550                 d_mark_dontcache(inode);
 551 }
 552
 553 static bool dax_compatible(struct inode *inode, unsigned int oldflags,
 554                            unsigned int flags)
 555 {
 556         /* Allow the DAX flag to be changed on inline directories */
 557         if (S_ISDIR(inode->i_mode)) {
 558                 flags &= ~EXT4_INLINE_DATA_FL;
 559                 oldflags &= ~EXT4_INLINE_DATA_FL;
 560         }
 561
 562         if (flags & EXT4_DAX_FL) {
 563                 if ((oldflags & EXT4_DAX_MUT_EXCL) ||
 564                      ext4_test_inode_state(inode,
 565                                           EXT4_STATE_VERITY_IN_PROGRESS)) {
 566                         return false;
 567                 }
 568         }
 569
 570         if ((flags & EXT4_DAX_MUT_EXCL) && (oldflags & EXT4_DAX_FL))
 571                         return false;
 572
 573         return true;
 574 }
 575
 576 static int ext4_ioctl_setflags(struct inode *inode,
 577                                unsigned int flags)
 578 {
 579         struct ext4_inode_info *ei = EXT4_I(inode);
 580         handle_t *handle = NULL;
 581         int err = -EPERM, migrate = 0;
 582         struct ext4_iloc iloc;
 583         unsigned int oldflags, mask, i;
 584         struct super_block *sb = inode->i_sb;
 585
 586         /* Is it quota file? Do not allow user to mess with it */
 587         if (ext4_is_quota_file(inode))
 588                 goto flags_out;
 589
 590         oldflags = ei->i_flags;
 591         /*
 592          * The JOURNAL_DATA flag can only be changed by
 593          * the relevant capability.
 594          */
 595         if ((flags ^ oldflags) & (EXT4_JOURNAL_DATA_FL)) {
 596                 if (!capable(CAP_SYS_RESOURCE))
 597                         goto flags_out;
 598         }
 599
 600         if (!dax_compatible(inode, oldflags, flags)) {
 601                 err = -EOPNOTSUPP;
 602                 goto flags_out;
 603         }
 604
 605         if ((flags ^ oldflags) & EXT4_EXTENTS_FL)
 606                 migrate = 1;
 607
 608         if ((flags ^ oldflags) & EXT4_CASEFOLD_FL) {
 609                 if (!ext4_has_feature_casefold(sb)) {
 610                         err = -EOPNOTSUPP;
 611                         goto flags_out;
 612                 }
 613
 614                 if (!S_ISDIR(inode->i_mode)) {
 615                         err = -ENOTDIR;
 616                         goto flags_out;
 617                 }
 618
 619                 if (!ext4_empty_dir(inode)) {
 620                         err = -ENOTEMPTY;
 621                         goto flags_out;
 622                 }
 623         }
 624
 625         /*
 626          * Wait for all pending directio and then flush all the dirty pages
 627          * for this file.  The flush marks all the pages readonly, so any
 628          * subsequent attempt to write to the file (particularly mmap pages)
 629          * will come through the filesystem and fail.
 630          */
 631         if (S_ISREG(inode->i_mode) && !IS_IMMUTABLE(inode) &&
 632             (flags & EXT4_IMMUTABLE_FL)) {
 633                 inode_dio_wait(inode);
 634                 err = filemap_write_and_wait(inode->i_mapping);
 635                 if (err)
 636                         goto flags_out;
 637         }
 638
 639         handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
 640         if (IS_ERR(handle)) {
 641                 err = PTR_ERR(handle);
 642                 goto flags_out;
 643         }
 644         if (IS_SYNC(inode))
 645                 ext4_handle_sync(handle);
 646         err = ext4_reserve_inode_write(handle, inode, &iloc);
 647         if (err)
 648                 goto flags_err;
 649
 650         ext4_dax_dontcache(inode, flags);
 651
 652         for (i = 0, mask = 1; i < 32; i++, mask <<= 1) {
 653                 if (!(mask & EXT4_FL_USER_MODIFIABLE))
 654                         continue;
 655                 /* These flags get special treatment later */
 656                 if (mask == EXT4_JOURNAL_DATA_FL || mask == EXT4_EXTENTS_FL)
 657                         continue;
 658                 if (mask & flags)
 659                         ext4_set_inode_flag(inode, i);
 660                 else
 661                         ext4_clear_inode_flag(inode, i);
 662         }
 663
 664         ext4_set_inode_flags(inode, false);
 665
 666         inode->i_ctime = current_time(inode);
 667         inode_inc_iversion(inode);
 668
 669         err = ext4_mark_iloc_dirty(handle, inode, &iloc);
 670 flags_err:
 671         ext4_journal_stop(handle);
 672         if (err)
 673                 goto flags_out;
 674
 675         if ((flags ^ oldflags) & (EXT4_JOURNAL_DATA_FL)) {
 676                 /*
 677                  * Changes to the journaling mode can cause unsafe changes to
 678                  * S_DAX if the inode is DAX
 679                  */
 680                 if (IS_DAX(inode)) {
 681                         err = -EBUSY;
 682                         goto flags_out;
 683                 }
 684
 685                 err = ext4_change_inode_journal_flag(inode,
 686                                                      flags & EXT4_JOURNAL_DATA_FL);
 687                 if (err)
 688                         goto flags_out;
 689         }
 690         if (migrate) {
 691                 if (flags & EXT4_EXTENTS_FL)
 692                         err = ext4_ext_migrate(inode);
 693                 else
 694                         err = ext4_ind_migrate(inode);
 695         }
 696
 697 flags_out:
 698         return err;
 699 }
 700
 701 #ifdef CONFIG_QUOTA
 702 static int ext4_ioctl_setproject(struct inode *inode, __u32 projid)
 703 {
 704         struct super_block *sb = inode->i_sb;
 705         struct ext4_inode_info *ei = EXT4_I(inode);
 706         int err, rc;
 707         handle_t *handle;
 708         kprojid_t kprojid;
 709         struct ext4_iloc iloc;
 710         struct ext4_inode *raw_inode;
 711         struct dquot *transfer_to[MAXQUOTAS] = { };
 712
 713         if (!ext4_has_feature_project(sb)) {
 714                 if (projid != EXT4_DEF_PROJID)
 715                         return -EOPNOTSUPP;
 716                 else
 717                         return 0;
 718         }
 719
 720         if (EXT4_INODE_SIZE(sb) <= EXT4_GOOD_OLD_INODE_SIZE)
 721                 return -EOPNOTSUPP;
 722
 723         kprojid = make_kprojid(&init_user_ns, (projid_t)projid);
 724
 725         if (projid_eq(kprojid, EXT4_I(inode)->i_projid))
 726                 return 0;
 727
 728         err = -EPERM;
 729         /* Is it quota file? Do not allow user to mess with it */
 730         if (ext4_is_quota_file(inode))
 731                 return err;
 732
 733         err = dquot_initialize(inode);
 734         if (err)
 735                 return err;
 736
 737         err = ext4_get_inode_loc(inode, &iloc);
 738         if (err)
 739                 return err;
 740
 741         raw_inode = ext4_raw_inode(&iloc);
 742         if (!EXT4_FITS_IN_INODE(raw_inode, ei, i_projid)) {
 743                 err = ext4_expand_extra_isize(inode,
 744                                               EXT4_SB(sb)->s_want_extra_isize,
 745                                               &iloc);
 746                 if (err)
 747                         return err;
 748         } else {
 749                 brelse(iloc.bh);
 750         }
 751
 752         handle = ext4_journal_start(inode, EXT4_HT_QUOTA,
 753                 EXT4_QUOTA_INIT_BLOCKS(sb) +
 754                 EXT4_QUOTA_DEL_BLOCKS(sb) + 3);
 755         if (IS_ERR(handle))
 756                 return PTR_ERR(handle);
 757
 758         err = ext4_reserve_inode_write(handle, inode, &iloc);
 759         if (err)
 760                 goto out_stop;
 761
 762         transfer_to[PRJQUOTA] = dqget(sb, make_kqid_projid(kprojid));
 763         if (!IS_ERR(transfer_to[PRJQUOTA])) {
 764
 765                 /* __dquot_transfer() calls back ext4_get_inode_usage() which
 766                  * counts xattr inode references.
 767                  */
 768                 down_read(&EXT4_I(inode)->xattr_sem);
 769                 err = __dquot_transfer(inode, transfer_to);
 770                 up_read(&EXT4_I(inode)->xattr_sem);
 771                 dqput(transfer_to[PRJQUOTA]);
 772                 if (err)
 773                         goto out_dirty;
 774         }
 775
 776         EXT4_I(inode)->i_projid = kprojid;
 777         inode->i_ctime = current_time(inode);
 778         inode_inc_iversion(inode);
 779 out_dirty:
 780         rc = ext4_mark_iloc_dirty(handle, inode, &iloc);
 781         if (!err)
 782                 err = rc;
 783 out_stop:
 784         ext4_journal_stop(handle);
 785         return err;
 786 }
 787 #else
 788 static int ext4_ioctl_setproject(struct inode *inode, __u32 projid)
 789 {
 790         if (projid != EXT4_DEF_PROJID)
 791                 return -EOPNOTSUPP;
 792         return 0;
 793 }
 794 #endif
 795
 796 int ext4_force_shutdown(struct super_block *sb, u32 flags)
 797 {
 798         struct ext4_sb_info *sbi = EXT4_SB(sb);
 799         int ret;
 800
 801         if (flags > EXT4_GOING_FLAGS_NOLOGFLUSH)
 802                 return -EINVAL;
 803
 804         if (ext4_forced_shutdown(sbi))
 805                 return 0;
 806
 807         ext4_msg(sb, KERN_ALERT, "shut down requested (%d)", flags);
 808         trace_ext4_shutdown(sb, flags);
 809
 810         switch (flags) {
 811         case EXT4_GOING_FLAGS_DEFAULT:
 812                 ret = freeze_bdev(sb->s_bdev);
 813                 if (ret)
 814                         return ret;
 815                 set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
 816                 thaw_bdev(sb->s_bdev);
 817                 break;
 818         case EXT4_GOING_FLAGS_LOGFLUSH:
 819                 set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
 820                 if (sbi->s_journal && !is_journal_aborted(sbi->s_journal)) {
 821                         (void) ext4_force_commit(sb);
 822                         jbd2_journal_abort(sbi->s_journal, -ESHUTDOWN);
 823                 }
 824                 break;
 825         case EXT4_GOING_FLAGS_NOLOGFLUSH:
 826                 set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
 827                 if (sbi->s_journal && !is_journal_aborted(sbi->s_journal))
 828                         jbd2_journal_abort(sbi->s_journal, -ESHUTDOWN);
 829                 break;
 830         default:
 831                 return -EINVAL;
 832         }
 833         clear_opt(sb, DISCARD);
 834         return 0;
 835 }
 836
 837 static int ext4_ioctl_shutdown(struct super_block *sb, unsigned long arg)
 838 {
 839         u32 flags;
 840
 841         if (!capable(CAP_SYS_ADMIN))
 842                 return -EPERM;
 843
 844         if (get_user(flags, (__u32 __user *)arg))
 845                 return -EFAULT;
 846
 847         return ext4_force_shutdown(sb, flags);
 848 }
 849
 850 struct getfsmap_info {
 851         struct super_block      *gi_sb;
 852         struct fsmap_head __user *gi_data;
 853         unsigned int            gi_idx;
 854         __u32                   gi_last_flags;
 855 };
 856
 857 static int ext4_getfsmap_format(struct ext4_fsmap *xfm, void *priv)
 858 {
 859         struct getfsmap_info *info = priv;
 860         struct fsmap fm;
 861
 862         trace_ext4_getfsmap_mapping(info->gi_sb, xfm);
 863
 864         info->gi_last_flags = xfm->fmr_flags;
 865         ext4_fsmap_from_internal(info->gi_sb, &fm, xfm);
 866         if (copy_to_user(&info->gi_data->fmh_recs[info->gi_idx++], &fm,
 867                         sizeof(struct fsmap)))
 868                 return -EFAULT;
 869
 870         return 0;
 871 }
 872
 873 static int ext4_ioc_getfsmap(struct super_block *sb,
 874                              struct fsmap_head __user *arg)
 875 {
 876         struct getfsmap_info info = { NULL };
 877         struct ext4_fsmap_head xhead = {0};
 878         struct fsmap_head head;
 879         bool aborted = false;
 880         int error;
 881
 882         if (copy_from_user(&head, arg, sizeof(struct fsmap_head)))
 883                 return -EFAULT;
 884         if (memchr_inv(head.fmh_reserved, 0, sizeof(head.fmh_reserved)) ||
 885             memchr_inv(head.fmh_keys[0].fmr_reserved, 0,
 886                        sizeof(head.fmh_keys[0].fmr_reserved)) ||
 887             memchr_inv(head.fmh_keys[1].fmr_reserved, 0,
 888                        sizeof(head.fmh_keys[1].fmr_reserved)))
 889                 return -EINVAL;
 890         /*
 891          * ext4 doesn't report file extents at all, so the only valid
 892          * file offsets are the magic ones (all zeroes or all ones).
 893          */
 894         if (head.fmh_keys[0].fmr_offset ||
 895             (head.fmh_keys[1].fmr_offset != 0 &&
 896              head.fmh_keys[1].fmr_offset != -1ULL))
 897                 return -EINVAL;
 898
 899         xhead.fmh_iflags = head.fmh_iflags;
 900         xhead.fmh_count = head.fmh_count;
 901         ext4_fsmap_to_internal(sb, &xhead.fmh_keys[0], &head.fmh_keys[0]);
 902         ext4_fsmap_to_internal(sb, &xhead.fmh_keys[1], &head.fmh_keys[1]);
 903
 904         trace_ext4_getfsmap_low_key(sb, &xhead.fmh_keys[0]);
 905         trace_ext4_getfsmap_high_key(sb, &xhead.fmh_keys[1]);
 906
 907         info.gi_sb = sb;
 908         info.gi_data = arg;
 909         error = ext4_getfsmap(sb, &xhead, ext4_getfsmap_format, &info);
 910         if (error == EXT4_QUERY_RANGE_ABORT)
 911                 aborted = true;
 912         else if (error)
 913                 return error;
 914
 915         /* If we didn't abort, set the "last" flag in the last fmx */
 916         if (!aborted && info.gi_idx) {
 917                 info.gi_last_flags |= FMR_OF_LAST;
 918                 if (copy_to_user(&info.gi_data->fmh_recs[info.gi_idx - 1].fmr_flags,
 919                                  &info.gi_last_flags,
 920                                  sizeof(info.gi_last_flags)))
 921                         return -EFAULT;
 922         }
 923
 924         /* copy back header */
 925         head.fmh_entries = xhead.fmh_entries;
 926         head.fmh_oflags = xhead.fmh_oflags;
 927         if (copy_to_user(arg, &head, sizeof(struct fsmap_head)))
 928                 return -EFAULT;
 929
 930         return 0;
 931 }
 932
 933 static long ext4_ioctl_group_add(struct file *file,
 934                                  struct ext4_new_group_data *input)
 935 {
 936         struct super_block *sb = file_inode(file)->i_sb;
 937         int err, err2=0;
 938
 939         err = ext4_resize_begin(sb);
 940         if (err)
 941                 return err;
 942
 943         if (ext4_has_feature_bigalloc(sb)) {
 944                 ext4_msg(sb, KERN_ERR,
 945                          "Online resizing not supported with bigalloc");
 946                 err = -EOPNOTSUPP;
 947                 goto group_add_out;
 948         }
 949
 950         err = mnt_want_write_file(file);
 951         if (err)
 952                 goto group_add_out;
 953
 954         err = ext4_group_add(sb, input);
 955         if (EXT4_SB(sb)->s_journal) {
 956                 jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
 957                 err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
 958                 jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
 959         }
 960         if (err == 0)
 961                 err = err2;
 962         mnt_drop_write_file(file);
 963         if (!err && ext4_has_group_desc_csum(sb) &&
 964             test_opt(sb, INIT_INODE_TABLE))
 965                 err = ext4_register_li_request(sb, input->group);
 966 group_add_out:
 967         err2 = ext4_resize_end(sb, false);
 968         if (err == 0)
 969                 err = err2;
 970         return err;
 971 }
 972
 973 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa)
 974 {
 975         struct inode *inode = d_inode(dentry);
 976         struct ext4_inode_info *ei = EXT4_I(inode);
 977         u32 flags = ei->i_flags & EXT4_FL_USER_VISIBLE;
 978
 979         if (S_ISREG(inode->i_mode))
 980                 flags &= ~FS_PROJINHERIT_FL;
 981
 982         fileattr_fill_flags(fa, flags);
 983         if (ext4_has_feature_project(inode->i_sb))
 984                 fa->fsx_projid = from_kprojid(&init_user_ns, ei->i_projid);
 985
 986         return 0;
 987 }
 988
 989 int ext4_fileattr_set(struct mnt_idmap *idmap,
 990                       struct dentry *dentry, struct fileattr *fa)
 991 {
 992         struct inode *inode = d_inode(dentry);
 993         u32 flags = fa->flags;
 994         int err = -EOPNOTSUPP;
 995
 996         if (flags & ~EXT4_FL_USER_VISIBLE)
 997                 goto out;
 998
 999         /*
1000          * chattr(1) grabs flags via GETFLAGS, modifies the result and
1001          * passes that to SETFLAGS. So we cannot easily make SETFLAGS
1002          * more restrictive than just silently masking off visible but
1003          * not settable flags as we always did.
1004          */
1005         flags &= EXT4_FL_USER_MODIFIABLE;
1006         if (ext4_mask_flags(inode->i_mode, flags) != flags)
1007                 goto out;
1008         err = ext4_ioctl_check_immutable(inode, fa->fsx_projid, flags);
1009         if (err)
1010                 goto out;
1011         err = ext4_ioctl_setflags(inode, flags);
1012         if (err)
1013                 goto out;
1014         err = ext4_ioctl_setproject(inode, fa->fsx_projid);
1015 out:
1016         return err;
1017 }
1018
1019 /* So that the fiemap access checks can't overflow on 32 bit machines. */
1020 #define FIEMAP_MAX_EXTENTS      (UINT_MAX / sizeof(struct fiemap_extent))
1021
1022 static int ext4_ioctl_get_es_cache(struct file *filp, unsigned long arg)
1023 {
1024         struct fiemap fiemap;
1025         struct fiemap __user *ufiemap = (struct fiemap __user *) arg;
1026         struct fiemap_extent_info fieinfo = { 0, };
1027         struct inode *inode = file_inode(filp);
1028         int error;
1029
1030         if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap)))
1031                 return -EFAULT;
1032
1033         if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
1034                 return -EINVAL;
1035
1036         fieinfo.fi_flags = fiemap.fm_flags;
1037         fieinfo.fi_extents_max = fiemap.fm_extent_count;
1038         fieinfo.fi_extents_start = ufiemap->fm_extents;
1039
1040         error = ext4_get_es_cache(inode, &fieinfo, fiemap.fm_start,
1041                         fiemap.fm_length);
1042         fiemap.fm_flags = fieinfo.fi_flags;
1043         fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
1044         if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap)))
1045                 error = -EFAULT;
1046
1047         return error;
1048 }
1049
1050 static int ext4_ioctl_checkpoint(struct file *filp, unsigned long arg)
1051 {
1052         int err = 0;
1053         __u32 flags = 0;
1054         unsigned int flush_flags = 0;
1055         struct super_block *sb = file_inode(filp)->i_sb;
1056
1057         if (copy_from_user(&flags, (__u32 __user *)arg,
1058                                 sizeof(__u32)))
1059                 return -EFAULT;
1060
1061         if (!capable(CAP_SYS_ADMIN))
1062                 return -EPERM;
1063
1064         /* check for invalid bits set */
1065         if ((flags & ~EXT4_IOC_CHECKPOINT_FLAG_VALID) ||
1066                                 ((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
1067                                 (flags & JBD2_JOURNAL_FLUSH_ZEROOUT)))
1068                 return -EINVAL;
1069
1070         if (!EXT4_SB(sb)->s_journal)
1071                 return -ENODEV;
1072
1073         if ((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
1074             !bdev_max_discard_sectors(EXT4_SB(sb)->s_journal->j_dev))
1075                 return -EOPNOTSUPP;
1076
1077         if (flags & EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
1078                 return 0;
1079
1080         if (flags & EXT4_IOC_CHECKPOINT_FLAG_DISCARD)
1081                 flush_flags |= JBD2_JOURNAL_FLUSH_DISCARD;
1082
1083         if (flags & EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT) {
1084                 flush_flags |= JBD2_JOURNAL_FLUSH_ZEROOUT;
1085                 pr_info_ratelimited("warning: checkpointing journal with EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT can be slow");
1086         }
1087
1088         jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1089         err = jbd2_journal_flush(EXT4_SB(sb)->s_journal, flush_flags);
1090         jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1091
1092         return err;
1093 }
1094
1095 static int ext4_ioctl_setlabel(struct file *filp, const char __user *user_label)
1096 {
1097         size_t len;
1098         int ret = 0;
1099         char new_label[EXT4_LABEL_MAX + 1];
1100         struct super_block *sb = file_inode(filp)->i_sb;
1101
1102         if (!capable(CAP_SYS_ADMIN))
1103                 return -EPERM;
1104
1105         /*
1106          * Copy the maximum length allowed for ext4 label with one more to
1107          * find the required terminating null byte in order to test the
1108          * label length. The on disk label doesn't need to be null terminated.
1109          */
1110         if (copy_from_user(new_label, user_label, EXT4_LABEL_MAX + 1))
1111                 return -EFAULT;
1112
1113         len = strnlen(new_label, EXT4_LABEL_MAX + 1);
1114         if (len > EXT4_LABEL_MAX)
1115                 return -EINVAL;
1116
1117         /*
1118          * Clear the buffer after the new label
1119          */
1120         memset(new_label + len, 0, EXT4_LABEL_MAX - len);
1121
1122         ret = mnt_want_write_file(filp);
1123         if (ret)
1124                 return ret;
1125
1126         ret = ext4_update_superblocks_fn(sb, ext4_sb_setlabel, new_label);
1127
1128         mnt_drop_write_file(filp);
1129         return ret;
1130 }
1131
1132 static int ext4_ioctl_getlabel(struct ext4_sb_info *sbi, char __user *user_label)
1133 {
1134         char label[EXT4_LABEL_MAX + 1];
1135
1136         /*
1137          * EXT4_LABEL_MAX must always be smaller than FSLABEL_MAX because
1138          * FSLABEL_MAX must include terminating null byte, while s_volume_name
1139          * does not have to.
1140          */
1141         BUILD_BUG_ON(EXT4_LABEL_MAX >= FSLABEL_MAX);
1142
1143         memset(label, 0, sizeof(label));
1144         lock_buffer(sbi->s_sbh);
1145         strncpy(label, sbi->s_es->s_volume_name, EXT4_LABEL_MAX);
1146         unlock_buffer(sbi->s_sbh);
1147
1148         if (copy_to_user(user_label, label, sizeof(label)))
1149                 return -EFAULT;
1150         return 0;
1151 }
1152
1153 static int ext4_ioctl_getuuid(struct ext4_sb_info *sbi,
1154                         struct fsuuid __user *ufsuuid)
1155 {
1156         struct fsuuid fsuuid;
1157         __u8 uuid[UUID_SIZE];
1158
1159         if (copy_from_user(&fsuuid, ufsuuid, sizeof(fsuuid)))
1160                 return -EFAULT;
1161
1162         if (fsuuid.fsu_len == 0) {
1163                 fsuuid.fsu_len = UUID_SIZE;
1164                 if (copy_to_user(&ufsuuid->fsu_len, &fsuuid.fsu_len,
1165                                         sizeof(fsuuid.fsu_len)))
1166                         return -EFAULT;
1167                 return 0;
1168         }
1169
1170         if (fsuuid.fsu_len < UUID_SIZE || fsuuid.fsu_flags != 0)
1171                 return -EINVAL;
1172
1173         lock_buffer(sbi->s_sbh);
1174         memcpy(uuid, sbi->s_es->s_uuid, UUID_SIZE);
1175         unlock_buffer(sbi->s_sbh);
1176
1177         fsuuid.fsu_len = UUID_SIZE;
1178         if (copy_to_user(ufsuuid, &fsuuid, sizeof(fsuuid)) ||
1179             copy_to_user(&ufsuuid->fsu_uuid[0], uuid, UUID_SIZE))
1180                 return -EFAULT;
1181         return 0;
1182 }
1183
1184 static int ext4_ioctl_setuuid(struct file *filp,
1185                         const struct fsuuid __user *ufsuuid)
1186 {
1187         int ret = 0;
1188         struct super_block *sb = file_inode(filp)->i_sb;
1189         struct fsuuid fsuuid;
1190         __u8 uuid[UUID_SIZE];
1191
1192         if (!capable(CAP_SYS_ADMIN))
1193                 return -EPERM;
1194
1195         /*
1196          * If any checksums (group descriptors or metadata) are being used
1197          * then the checksum seed feature is required to change the UUID.
1198          */
1199         if (((ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb))
1200                         && !ext4_has_feature_csum_seed(sb))
1201                 || ext4_has_feature_stable_inodes(sb))
1202                 return -EOPNOTSUPP;
1203
1204         if (copy_from_user(&fsuuid, ufsuuid, sizeof(fsuuid)))
1205                 return -EFAULT;
1206
1207         if (fsuuid.fsu_len != UUID_SIZE || fsuuid.fsu_flags != 0)
1208                 return -EINVAL;
1209
1210         if (copy_from_user(uuid, &ufsuuid->fsu_uuid[0], UUID_SIZE))
1211                 return -EFAULT;
1212
1213         ret = mnt_want_write_file(filp);
1214         if (ret)
1215                 return ret;
1216
1217         ret = ext4_update_superblocks_fn(sb, ext4_sb_setuuid, &uuid);
1218         mnt_drop_write_file(filp);
1219
1220         return ret;
1221 }
1222
1223 static long __ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
1224 {
1225         struct inode *inode = file_inode(filp);
1226         struct super_block *sb = inode->i_sb;
1227         struct mnt_idmap *idmap = file_mnt_idmap(filp);
1228
1229         ext4_debug("cmd = %u, arg = %lu\n", cmd, arg);
1230
1231         switch (cmd) {
1232         case FS_IOC_GETFSMAP:
1233                 return ext4_ioc_getfsmap(sb, (void __user *)arg);
1234         case EXT4_IOC_GETVERSION:
1235         case EXT4_IOC_GETVERSION_OLD:
1236                 return put_user(inode->i_generation, (int __user *) arg);
1237         case EXT4_IOC_SETVERSION:
1238         case EXT4_IOC_SETVERSION_OLD: {
1239                 handle_t *handle;
1240                 struct ext4_iloc iloc;
1241                 __u32 generation;
1242                 int err;
1243
1244                 if (!inode_owner_or_capable(idmap, inode))
1245                         return -EPERM;
1246
1247                 if (ext4_has_metadata_csum(inode->i_sb)) {
1248                         ext4_warning(sb, "Setting inode version is not "
1249                                      "supported with metadata_csum enabled.");
1250                         return -ENOTTY;
1251                 }
1252
1253                 err = mnt_want_write_file(filp);
1254                 if (err)
1255                         return err;
1256                 if (get_user(generation, (int __user *) arg)) {
1257                         err = -EFAULT;
1258                         goto setversion_out;
1259                 }
1260
1261                 inode_lock(inode);
1262                 handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
1263                 if (IS_ERR(handle)) {
1264                         err = PTR_ERR(handle);
1265                         goto unlock_out;
1266                 }
1267                 err = ext4_reserve_inode_write(handle, inode, &iloc);
1268                 if (err == 0) {
1269                         inode->i_ctime = current_time(inode);
1270                         inode_inc_iversion(inode);
1271                         inode->i_generation = generation;
1272                         err = ext4_mark_iloc_dirty(handle, inode, &iloc);
1273                 }
1274                 ext4_journal_stop(handle);
1275
1276 unlock_out:
1277                 inode_unlock(inode);
1278 setversion_out:
1279                 mnt_drop_write_file(filp);
1280                 return err;
1281         }
1282         case EXT4_IOC_GROUP_EXTEND: {
1283                 ext4_fsblk_t n_blocks_count;
1284                 int err, err2=0;
1285
1286                 err = ext4_resize_begin(sb);
1287                 if (err)
1288                         return err;
1289
1290                 if (get_user(n_blocks_count, (__u32 __user *)arg)) {
1291                         err = -EFAULT;
1292                         goto group_extend_out;
1293                 }
1294
1295                 if (ext4_has_feature_bigalloc(sb)) {
1296                         ext4_msg(sb, KERN_ERR,
1297                                  "Online resizing not supported with bigalloc");
1298                         err = -EOPNOTSUPP;
1299                         goto group_extend_out;
1300                 }
1301
1302                 err = mnt_want_write_file(filp);
1303                 if (err)
1304                         goto group_extend_out;
1305
1306                 err = ext4_group_extend(sb, EXT4_SB(sb)->s_es, n_blocks_count);
1307                 if (EXT4_SB(sb)->s_journal) {
1308                         jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1309                         err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
1310                         jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1311                 }
1312                 if (err == 0)
1313                         err = err2;
1314                 mnt_drop_write_file(filp);
1315 group_extend_out:
1316                 err2 = ext4_resize_end(sb, false);
1317                 if (err == 0)
1318                         err = err2;
1319                 return err;
1320         }
1321
1322         case EXT4_IOC_MOVE_EXT: {
1323                 struct move_extent me;
1324                 struct fd donor;
1325                 int err;
1326
1327                 if (!(filp->f_mode & FMODE_READ) ||
1328                     !(filp->f_mode & FMODE_WRITE))
1329                         return -EBADF;
1330
1331                 if (copy_from_user(&me,
1332                         (struct move_extent __user *)arg, sizeof(me)))
1333                         return -EFAULT;
1334                 me.moved_len = 0;
1335
1336                 donor = fdget(me.donor_fd);
1337                 if (!donor.file)
1338                         return -EBADF;
1339
1340                 if (!(donor.file->f_mode & FMODE_WRITE)) {
1341                         err = -EBADF;
1342                         goto mext_out;
1343                 }
1344
1345                 if (ext4_has_feature_bigalloc(sb)) {
1346                         ext4_msg(sb, KERN_ERR,
1347                                  "Online defrag not supported with bigalloc");
1348                         err = -EOPNOTSUPP;
1349                         goto mext_out;
1350                 } else if (IS_DAX(inode)) {
1351                         ext4_msg(sb, KERN_ERR,
1352                                  "Online defrag not supported with DAX");
1353                         err = -EOPNOTSUPP;
1354                         goto mext_out;
1355                 }
1356
1357                 err = mnt_want_write_file(filp);
1358                 if (err)
1359                         goto mext_out;
1360
1361                 err = ext4_move_extents(filp, donor.file, me.orig_start,
1362                                         me.donor_start, me.len, &me.moved_len);
1363                 mnt_drop_write_file(filp);
1364
1365                 if (copy_to_user((struct move_extent __user *)arg,
1366                                  &me, sizeof(me)))
1367                         err = -EFAULT;
1368 mext_out:
1369                 fdput(donor);
1370                 return err;
1371         }
1372
1373         case EXT4_IOC_GROUP_ADD: {
1374                 struct ext4_new_group_data input;
1375
1376                 if (copy_from_user(&input, (struct ext4_new_group_input __user *)arg,
1377                                 sizeof(input)))
1378                         return -EFAULT;
1379
1380                 return ext4_ioctl_group_add(filp, &input);
1381         }
1382
1383         case EXT4_IOC_MIGRATE:
1384         {
1385                 int err;
1386                 if (!inode_owner_or_capable(idmap, inode))
1387                         return -EACCES;
1388
1389                 err = mnt_want_write_file(filp);
1390                 if (err)
1391                         return err;
1392                 /*
1393                  * inode_mutex prevent write and truncate on the file.
1394                  * Read still goes through. We take i_data_sem in
1395                  * ext4_ext_swap_inode_data before we switch the
1396                  * inode format to prevent read.
1397                  */
1398                 inode_lock((inode));
1399                 err = ext4_ext_migrate(inode);
1400                 inode_unlock((inode));
1401                 mnt_drop_write_file(filp);
1402                 return err;
1403         }
1404
1405         case EXT4_IOC_ALLOC_DA_BLKS:
1406         {
1407                 int err;
1408                 if (!inode_owner_or_capable(idmap, inode))
1409                         return -EACCES;
1410
1411                 err = mnt_want_write_file(filp);
1412                 if (err)
1413                         return err;
1414                 err = ext4_alloc_da_blocks(inode);
1415                 mnt_drop_write_file(filp);
1416                 return err;
1417         }
1418
1419         case EXT4_IOC_SWAP_BOOT:
1420         {
1421                 int err;
1422                 if (!(filp->f_mode & FMODE_WRITE))
1423                         return -EBADF;
1424                 err = mnt_want_write_file(filp);
1425                 if (err)
1426                         return err;
1427                 err = swap_inode_boot_loader(sb, idmap, inode);
1428                 mnt_drop_write_file(filp);
1429                 return err;
1430         }
1431
1432         case EXT4_IOC_RESIZE_FS: {
1433                 ext4_fsblk_t n_blocks_count;
1434                 int err = 0, err2 = 0;
1435                 ext4_group_t o_group = EXT4_SB(sb)->s_groups_count;
1436
1437                 if (copy_from_user(&n_blocks_count, (__u64 __user *)arg,
1438                                    sizeof(__u64))) {
1439                         return -EFAULT;
1440                 }
1441
1442                 err = ext4_resize_begin(sb);
1443                 if (err)
1444                         return err;
1445
1446                 err = mnt_want_write_file(filp);
1447                 if (err)
1448                         goto resizefs_out;
1449
1450                 err = ext4_resize_fs(sb, n_blocks_count);
1451                 if (EXT4_SB(sb)->s_journal) {
1452                         ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_RESIZE, NULL);
1453                         jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
1454                         err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
1455                         jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
1456                 }
1457                 if (err == 0)
1458                         err = err2;
1459                 mnt_drop_write_file(filp);
1460                 if (!err && (o_group < EXT4_SB(sb)->s_groups_count) &&
1461                     ext4_has_group_desc_csum(sb) &&
1462                     test_opt(sb, INIT_INODE_TABLE))
1463                         err = ext4_register_li_request(sb, o_group);
1464
1465 resizefs_out:
1466                 err2 = ext4_resize_end(sb, true);
1467                 if (err == 0)
1468                         err = err2;
1469                 return err;
1470         }
1471
1472         case FITRIM:
1473         {
1474                 struct fstrim_range range;
1475                 int ret = 0;
1476
1477                 if (!capable(CAP_SYS_ADMIN))
1478                         return -EPERM;
1479
1480                 if (!bdev_max_discard_sectors(sb->s_bdev))
1481                         return -EOPNOTSUPP;
1482
1483                 /*
1484                  * We haven't replayed the journal, so we cannot use our
1485                  * block-bitmap-guided storage zapping commands.
1486                  */
1487                 if (test_opt(sb, NOLOAD) && ext4_has_feature_journal(sb))
1488                         return -EROFS;
1489
1490                 if (copy_from_user(&range, (struct fstrim_range __user *)arg,
1491                     sizeof(range)))
1492                         return -EFAULT;
1493
1494                 ret = ext4_trim_fs(sb, &range);
1495                 if (ret < 0)
1496                         return ret;
1497
1498                 if (copy_to_user((struct fstrim_range __user *)arg, &range,
1499                     sizeof(range)))
1500                         return -EFAULT;
1501
1502                 return 0;
1503         }
1504         case EXT4_IOC_PRECACHE_EXTENTS:
1505                 return ext4_ext_precache(inode);
1506
1507         case FS_IOC_SET_ENCRYPTION_POLICY:
1508                 if (!ext4_has_feature_encrypt(sb))
1509                         return -EOPNOTSUPP;
1510                 return fscrypt_ioctl_set_policy(filp, (const void __user *)arg);
1511
1512         case FS_IOC_GET_ENCRYPTION_PWSALT:
1513                 return ext4_ioctl_get_encryption_pwsalt(filp, (void __user *)arg);
1514
1515         case FS_IOC_GET_ENCRYPTION_POLICY:
1516                 if (!ext4_has_feature_encrypt(sb))
1517                         return -EOPNOTSUPP;
1518                 return fscrypt_ioctl_get_policy(filp, (void __user *)arg);
1519
1520         case FS_IOC_GET_ENCRYPTION_POLICY_EX:
1521                 if (!ext4_has_feature_encrypt(sb))
1522                         return -EOPNOTSUPP;
1523                 return fscrypt_ioctl_get_policy_ex(filp, (void __user *)arg);
1524
1525         case FS_IOC_ADD_ENCRYPTION_KEY:
1526                 if (!ext4_has_feature_encrypt(sb))
1527                         return -EOPNOTSUPP;
1528                 return fscrypt_ioctl_add_key(filp, (void __user *)arg);
1529
1530         case FS_IOC_REMOVE_ENCRYPTION_KEY:
1531                 if (!ext4_has_feature_encrypt(sb))
1532                         return -EOPNOTSUPP;
1533                 return fscrypt_ioctl_remove_key(filp, (void __user *)arg);
1534
1535         case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS:
1536                 if (!ext4_has_feature_encrypt(sb))
1537                         return -EOPNOTSUPP;
1538                 return fscrypt_ioctl_remove_key_all_users(filp,
1539                                                           (void __user *)arg);
1540         case FS_IOC_GET_ENCRYPTION_KEY_STATUS:
1541                 if (!ext4_has_feature_encrypt(sb))
1542                         return -EOPNOTSUPP;
1543                 return fscrypt_ioctl_get_key_status(filp, (void __user *)arg);
1544
1545         case FS_IOC_GET_ENCRYPTION_NONCE:
1546                 if (!ext4_has_feature_encrypt(sb))
1547                         return -EOPNOTSUPP;
1548                 return fscrypt_ioctl_get_nonce(filp, (void __user *)arg);
1549
1550         case EXT4_IOC_CLEAR_ES_CACHE:
1551         {
1552                 if (!inode_owner_or_capable(idmap, inode))
1553                         return -EACCES;
1554                 ext4_clear_inode_es(inode);
1555                 return 0;
1556         }
1557
1558         case EXT4_IOC_GETSTATE:
1559         {
1560                 __u32   state = 0;
1561
1562                 if (ext4_test_inode_state(inode, EXT4_STATE_EXT_PRECACHED))
1563                         state |= EXT4_STATE_FLAG_EXT_PRECACHED;
1564                 if (ext4_test_inode_state(inode, EXT4_STATE_NEW))
1565                         state |= EXT4_STATE_FLAG_NEW;
1566                 if (ext4_test_inode_state(inode, EXT4_STATE_NEWENTRY))
1567                         state |= EXT4_STATE_FLAG_NEWENTRY;
1568                 if (ext4_test_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE))
1569                         state |= EXT4_STATE_FLAG_DA_ALLOC_CLOSE;
1570
1571                 return put_user(state, (__u32 __user *) arg);
1572         }
1573
1574         case EXT4_IOC_GET_ES_CACHE:
1575                 return ext4_ioctl_get_es_cache(filp, arg);
1576
1577         case EXT4_IOC_SHUTDOWN:
1578                 return ext4_ioctl_shutdown(sb, arg);
1579
1580         case FS_IOC_ENABLE_VERITY:
1581                 if (!ext4_has_feature_verity(sb))
1582                         return -EOPNOTSUPP;
1583                 return fsverity_ioctl_enable(filp, (const void __user *)arg);
1584
1585         case FS_IOC_MEASURE_VERITY:
1586                 if (!ext4_has_feature_verity(sb))
1587                         return -EOPNOTSUPP;
1588                 return fsverity_ioctl_measure(filp, (void __user *)arg);
1589
1590         case FS_IOC_READ_VERITY_METADATA:
1591                 if (!ext4_has_feature_verity(sb))
1592                         return -EOPNOTSUPP;
1593                 return fsverity_ioctl_read_metadata(filp,
1594                                                     (const void __user *)arg);
1595
1596         case EXT4_IOC_CHECKPOINT:
1597                 return ext4_ioctl_checkpoint(filp, arg);
1598
1599         case FS_IOC_GETFSLABEL:
1600                 return ext4_ioctl_getlabel(EXT4_SB(sb), (void __user *)arg);
1601
1602         case FS_IOC_SETFSLABEL:
1603                 return ext4_ioctl_setlabel(filp,
1604                                            (const void __user *)arg);
1605
1606         case EXT4_IOC_GETFSUUID:
1607                 return ext4_ioctl_getuuid(EXT4_SB(sb), (void __user *)arg);
1608         case EXT4_IOC_SETFSUUID:
1609                 return ext4_ioctl_setuuid(filp, (const void __user *)arg);
1610         default:
1611                 return -ENOTTY;
1612         }
1613 }
1614
1615 long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
1616 {
1617         return __ext4_ioctl(filp, cmd, arg);
1618 }
1619
1620 #ifdef CONFIG_COMPAT
1621 long ext4_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
1622 {
1623         /* These are just misnamed, they actually get/put from/to user an int */
1624         switch (cmd) {
1625         case EXT4_IOC32_GETVERSION:
1626                 cmd = EXT4_IOC_GETVERSION;
1627                 break;
1628         case EXT4_IOC32_SETVERSION:
1629                 cmd = EXT4_IOC_SETVERSION;
1630                 break;
1631         case EXT4_IOC32_GROUP_EXTEND:
1632                 cmd = EXT4_IOC_GROUP_EXTEND;
1633                 break;
1634         case EXT4_IOC32_GETVERSION_OLD:
1635                 cmd = EXT4_IOC_GETVERSION_OLD;
1636                 break;
1637         case EXT4_IOC32_SETVERSION_OLD:
1638                 cmd = EXT4_IOC_SETVERSION_OLD;
1639                 break;
1640         case EXT4_IOC32_GETRSVSZ:
1641                 cmd = EXT4_IOC_GETRSVSZ;
1642                 break;
1643         case EXT4_IOC32_SETRSVSZ:
1644                 cmd = EXT4_IOC_SETRSVSZ;
1645                 break;
1646         case EXT4_IOC32_GROUP_ADD: {
1647                 struct compat_ext4_new_group_input __user *uinput;
1648                 struct ext4_new_group_data input;
1649                 int err;
1650
1651                 uinput = compat_ptr(arg);
1652                 err = get_user(input.group, &uinput->group);
1653                 err |= get_user(input.block_bitmap, &uinput->block_bitmap);
1654                 err |= get_user(input.inode_bitmap, &uinput->inode_bitmap);
1655                 err |= get_user(input.inode_table, &uinput->inode_table);
1656                 err |= get_user(input.blocks_count, &uinput->blocks_count);
1657                 err |= get_user(input.reserved_blocks,
1658                                 &uinput->reserved_blocks);
1659                 if (err)
1660                         return -EFAULT;
1661                 return ext4_ioctl_group_add(file, &input);
1662         }
1663         case EXT4_IOC_MOVE_EXT:
1664         case EXT4_IOC_RESIZE_FS:
1665         case FITRIM:
1666         case EXT4_IOC_PRECACHE_EXTENTS:
1667         case FS_IOC_SET_ENCRYPTION_POLICY:
1668         case FS_IOC_GET_ENCRYPTION_PWSALT:
1669         case FS_IOC_GET_ENCRYPTION_POLICY:
1670         case FS_IOC_GET_ENCRYPTION_POLICY_EX:
1671         case FS_IOC_ADD_ENCRYPTION_KEY:
1672         case FS_IOC_REMOVE_ENCRYPTION_KEY:
1673         case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS:
1674         case FS_IOC_GET_ENCRYPTION_KEY_STATUS:
1675         case FS_IOC_GET_ENCRYPTION_NONCE:
1676         case EXT4_IOC_SHUTDOWN:
1677         case FS_IOC_GETFSMAP:
1678         case FS_IOC_ENABLE_VERITY:
1679         case FS_IOC_MEASURE_VERITY:
1680         case FS_IOC_READ_VERITY_METADATA:
1681         case EXT4_IOC_CLEAR_ES_CACHE:
1682         case EXT4_IOC_GETSTATE:
1683         case EXT4_IOC_GET_ES_CACHE:
1684         case EXT4_IOC_CHECKPOINT:
1685         case FS_IOC_GETFSLABEL:
1686         case FS_IOC_SETFSLABEL:
1687         case EXT4_IOC_GETFSUUID:
1688         case EXT4_IOC_SETFSUUID:
1689                 break;
1690         default:
1691                 return -ENOIOCTLCMD;
1692         }
1693         return ext4_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
1694 }
1695 #endif
1696
1697 static void set_overhead(struct ext4_super_block *es, const void *arg)
1698 {
1699         es->s_overhead_clusters = cpu_to_le32(*((unsigned long *) arg));
1700 }
1701
1702 int ext4_update_overhead(struct super_block *sb, bool force)
1703 {
1704         struct ext4_sb_info *sbi = EXT4_SB(sb);
1705
1706         if (sb_rdonly(sb))
1707                 return 0;
1708         if (!force &&
1709             (sbi->s_overhead == 0 ||
1710              sbi->s_overhead == le32_to_cpu(sbi->s_es->s_overhead_clusters)))
1711                 return 0;
1712         return ext4_update_superblocks_fn(sb, set_overhead, &sbi->s_overhead);
1713 }