2 * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
4 * Copyright (c) 2001-2012 Anton Altaparmakov and Tuxera Inc.
5 * Copyright (c) 2001,2002 Richard Russon
7 * This program/include file is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License as published
9 * by the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * This program/include file is distributed in the hope that it will be
13 * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with this program (in the main directory of the Linux-NTFS
19 * distribution in the file COPYING); if not, write to the Free Software
20 * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
22 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
24 #include <linux/stddef.h>
25 #include <linux/init.h>
26 #include <linux/slab.h>
27 #include <linux/string.h>
28 #include <linux/spinlock.h>
29 #include <linux/blkdev.h> /* For bdev_logical_block_size(). */
30 #include <linux/backing-dev.h>
31 #include <linux/buffer_head.h>
32 #include <linux/vfs.h>
33 #include <linux/moduleparam.h>
34 #include <linux/bitmap.h>
49 /* Number of mounted filesystems which have compression enabled. */
50 static unsigned long ntfs_nr_compression_users;
52 /* A global default upcase table and a corresponding reference count. */
53 static ntfschar *default_upcase;
54 static unsigned long ntfs_nr_upcase_users;
56 /* Error constants/strings used in inode.c::ntfs_show_options(). */
58 /* One of these must be present, default is ON_ERRORS_CONTINUE. */
59 ON_ERRORS_PANIC = 0x01,
60 ON_ERRORS_REMOUNT_RO = 0x02,
61 ON_ERRORS_CONTINUE = 0x04,
62 /* Optional, can be combined with any of the above. */
63 ON_ERRORS_RECOVER = 0x10,
66 const option_t on_errors_arr[] = {
67 { ON_ERRORS_PANIC, "panic" },
68 { ON_ERRORS_REMOUNT_RO, "remount-ro", },
69 { ON_ERRORS_CONTINUE, "continue", },
70 { ON_ERRORS_RECOVER, "recover" },
77 * Copied from old ntfs driver (which copied from vfat driver).
79 static int simple_getbool(char *s, bool *setval)
82 if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
84 else if (!strcmp(s, "0") || !strcmp(s, "no") ||
95 * parse_options - parse the (re)mount options
97 * @opt: string containing the (re)mount options
99 * Parse the recognized options in @opt for the ntfs volume described by @vol.
101 static bool parse_options(ntfs_volume *vol, char *opt)
104 static char *utf8 = "utf8";
105 int errors = 0, sloppy = 0;
106 kuid_t uid = INVALID_UID;
107 kgid_t gid = INVALID_GID;
108 umode_t fmask = (umode_t)-1, dmask = (umode_t)-1;
109 int mft_zone_multiplier = -1, on_errors = -1;
110 int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
111 struct nls_table *nls_map = NULL, *old_nls;
113 /* I am lazy... (-8 */
114 #define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value) \
115 if (!strcmp(p, option)) { \
117 variable = default_value; \
119 variable = simple_strtoul(ov = v, &v, 0); \
124 #define NTFS_GETOPT(option, variable) \
125 if (!strcmp(p, option)) { \
128 variable = simple_strtoul(ov = v, &v, 0); \
132 #define NTFS_GETOPT_UID(option, variable) \
133 if (!strcmp(p, option)) { \
137 uid_value = simple_strtoul(ov = v, &v, 0); \
140 variable = make_kuid(current_user_ns(), uid_value); \
141 if (!uid_valid(variable)) \
144 #define NTFS_GETOPT_GID(option, variable) \
145 if (!strcmp(p, option)) { \
149 gid_value = simple_strtoul(ov = v, &v, 0); \
152 variable = make_kgid(current_user_ns(), gid_value); \
153 if (!gid_valid(variable)) \
156 #define NTFS_GETOPT_OCTAL(option, variable) \
157 if (!strcmp(p, option)) { \
160 variable = simple_strtoul(ov = v, &v, 8); \
164 #define NTFS_GETOPT_BOOL(option, variable) \
165 if (!strcmp(p, option)) { \
167 if (!simple_getbool(v, &val)) \
171 #define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array) \
172 if (!strcmp(p, option)) { \
177 if (variable == -1) \
179 for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
180 if (!strcmp(opt_array[_i].str, v)) { \
181 variable |= opt_array[_i].val; \
184 if (!opt_array[_i].str || !*opt_array[_i].str) \
188 goto no_mount_options;
189 ntfs_debug("Entering with mount options string: %s", opt);
190 while ((p = strsep(&opt, ","))) {
191 if ((v = strchr(p, '=')))
193 NTFS_GETOPT_UID("uid", uid)
194 else NTFS_GETOPT_GID("gid", gid)
195 else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
196 else NTFS_GETOPT_OCTAL("fmask", fmask)
197 else NTFS_GETOPT_OCTAL("dmask", dmask)
198 else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
199 else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
200 else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
201 else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
202 else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
203 else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
205 else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
206 ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
208 else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
209 if (!strcmp(p, "iocharset"))
210 ntfs_warning(vol->sb, "Option iocharset is "
211 "deprecated. Please use "
212 "option nls=<charsetname> in "
218 nls_map = load_nls(v);
221 ntfs_error(vol->sb, "NLS character set "
225 ntfs_error(vol->sb, "NLS character set %s not "
226 "found. Using previous one %s.",
227 v, old_nls->charset);
229 } else /* nls_map */ {
232 } else if (!strcmp(p, "utf8")) {
234 ntfs_warning(vol->sb, "Option utf8 is no longer "
235 "supported, using option nls=utf8. Please "
236 "use option nls=utf8 in the future and "
237 "make sure utf8 is compiled either as a "
238 "module or into the kernel.");
241 else if (!simple_getbool(v, &val))
248 ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
249 if (errors < INT_MAX)
252 #undef NTFS_GETOPT_OPTIONS_ARRAY
253 #undef NTFS_GETOPT_BOOL
255 #undef NTFS_GETOPT_WITH_DEFAULT
258 if (errors && !sloppy)
261 ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
262 "unrecognized mount option(s) and continuing.");
263 /* Keep this first! */
264 if (on_errors != -1) {
266 ntfs_error(vol->sb, "Invalid errors option argument "
267 "or bug in options parser.");
272 if (vol->nls_map && vol->nls_map != nls_map) {
273 ntfs_error(vol->sb, "Cannot change NLS character set "
276 } /* else (!vol->nls_map) */
277 ntfs_debug("Using NLS character set %s.", nls_map->charset);
278 vol->nls_map = nls_map;
279 } else /* (!nls_map) */ {
281 vol->nls_map = load_nls_default();
283 ntfs_error(vol->sb, "Failed to load default "
284 "NLS character set.");
287 ntfs_debug("Using default NLS character set (%s).",
288 vol->nls_map->charset);
291 if (mft_zone_multiplier != -1) {
292 if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
293 mft_zone_multiplier) {
294 ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
298 if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
299 ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
300 "Using default value, i.e. 1.");
301 mft_zone_multiplier = 1;
303 vol->mft_zone_multiplier = mft_zone_multiplier;
305 if (!vol->mft_zone_multiplier)
306 vol->mft_zone_multiplier = 1;
308 vol->on_errors = on_errors;
309 if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
310 vol->on_errors |= ON_ERRORS_CONTINUE;
315 if (fmask != (umode_t)-1)
317 if (dmask != (umode_t)-1)
319 if (show_sys_files != -1) {
321 NVolSetShowSystemFiles(vol);
323 NVolClearShowSystemFiles(vol);
325 if (case_sensitive != -1) {
327 NVolSetCaseSensitive(vol);
329 NVolClearCaseSensitive(vol);
331 if (disable_sparse != -1) {
333 NVolClearSparseEnabled(vol);
335 if (!NVolSparseEnabled(vol) &&
336 vol->major_ver && vol->major_ver < 3)
337 ntfs_warning(vol->sb, "Not enabling sparse "
338 "support due to NTFS volume "
339 "version %i.%i (need at least "
340 "version 3.0).", vol->major_ver,
343 NVolSetSparseEnabled(vol);
348 ntfs_error(vol->sb, "The %s option requires an argument.", p);
351 ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
354 ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
361 * ntfs_write_volume_flags - write new flags to the volume information flags
362 * @vol: ntfs volume on which to modify the flags
363 * @flags: new flags value for the volume information flags
365 * Internal function. You probably want to use ntfs_{set,clear}_volume_flags()
366 * instead (see below).
368 * Replace the volume information flags on the volume @vol with the value
369 * supplied in @flags. Note, this overwrites the volume information flags, so
370 * make sure to combine the flags you want to modify with the old flags and use
371 * the result when calling ntfs_write_volume_flags().
373 * Return 0 on success and -errno on error.
375 static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
377 ntfs_inode *ni = NTFS_I(vol->vol_ino);
379 VOLUME_INFORMATION *vi;
380 ntfs_attr_search_ctx *ctx;
383 ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
384 le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
385 if (vol->vol_flags == flags)
388 m = map_mft_record(ni);
393 ctx = ntfs_attr_get_search_ctx(ni, m);
396 goto put_unm_err_out;
398 err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
401 goto put_unm_err_out;
402 vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
403 le16_to_cpu(ctx->attr->data.resident.value_offset));
404 vol->vol_flags = vi->flags = flags;
405 flush_dcache_mft_record_page(ctx->ntfs_ino);
406 mark_mft_record_dirty(ctx->ntfs_ino);
407 ntfs_attr_put_search_ctx(ctx);
408 unmap_mft_record(ni);
414 ntfs_attr_put_search_ctx(ctx);
415 unmap_mft_record(ni);
417 ntfs_error(vol->sb, "Failed with error code %i.", -err);
422 * ntfs_set_volume_flags - set bits in the volume information flags
423 * @vol: ntfs volume on which to modify the flags
424 * @flags: flags to set on the volume
426 * Set the bits in @flags in the volume information flags on the volume @vol.
428 * Return 0 on success and -errno on error.
430 static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
432 flags &= VOLUME_FLAGS_MASK;
433 return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
437 * ntfs_clear_volume_flags - clear bits in the volume information flags
438 * @vol: ntfs volume on which to modify the flags
439 * @flags: flags to clear on the volume
441 * Clear the bits in @flags in the volume information flags on the volume @vol.
443 * Return 0 on success and -errno on error.
445 static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
447 flags &= VOLUME_FLAGS_MASK;
448 flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
449 return ntfs_write_volume_flags(vol, flags);
455 * ntfs_remount - change the mount options of a mounted ntfs filesystem
456 * @sb: superblock of mounted ntfs filesystem
457 * @flags: remount flags
458 * @opt: remount options string
460 * Change the mount options of an already mounted ntfs filesystem.
462 * NOTE: The VFS sets the @sb->s_flags remount flags to @flags after
463 * ntfs_remount() returns successfully (i.e. returns 0). Otherwise,
464 * @sb->s_flags are not changed.
466 static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
468 ntfs_volume *vol = NTFS_SB(sb);
470 ntfs_debug("Entering with remount options string: %s", opt);
475 /* For read-only compiled driver, enforce read-only flag. */
479 * For the read-write compiled driver, if we are remounting read-write,
480 * make sure there are no volume errors and that no unsupported volume
481 * flags are set. Also, empty the logfile journal as it would become
482 * stale as soon as something is written to the volume and mark the
483 * volume dirty so that chkdsk is run if the volume is not umounted
484 * cleanly. Finally, mark the quotas out of date so Windows rescans
485 * the volume on boot and updates them.
487 * When remounting read-only, mark the volume clean if no volume errors
490 if (sb_rdonly(sb) && !(*flags & SB_RDONLY)) {
491 static const char *es = ". Cannot remount read-write.";
493 /* Remounting read-write. */
494 if (NVolErrors(vol)) {
495 ntfs_error(sb, "Volume has errors and is read-only%s",
499 if (vol->vol_flags & VOLUME_IS_DIRTY) {
500 ntfs_error(sb, "Volume is dirty and read-only%s", es);
503 if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
504 ntfs_error(sb, "Volume has been modified by chkdsk "
505 "and is read-only%s", es);
508 if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
509 ntfs_error(sb, "Volume has unsupported flags set "
510 "(0x%x) and is read-only%s",
511 (unsigned)le16_to_cpu(vol->vol_flags),
515 if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
516 ntfs_error(sb, "Failed to set dirty bit in volume "
517 "information flags%s", es);
521 // TODO: Enable this code once we start modifying anything that
522 // is different between NTFS 1.2 and 3.x...
523 /* Set NT4 compatibility flag on newer NTFS version volumes. */
524 if ((vol->major_ver > 1)) {
525 if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
526 ntfs_error(sb, "Failed to set NT4 "
527 "compatibility flag%s", es);
533 if (!ntfs_empty_logfile(vol->logfile_ino)) {
534 ntfs_error(sb, "Failed to empty journal $LogFile%s",
539 if (!ntfs_mark_quotas_out_of_date(vol)) {
540 ntfs_error(sb, "Failed to mark quotas out of date%s",
545 if (!ntfs_stamp_usnjrnl(vol)) {
546 ntfs_error(sb, "Failed to stamp transaction log "
551 } else if (!sb_rdonly(sb) && (*flags & SB_RDONLY)) {
552 /* Remounting read-only. */
553 if (!NVolErrors(vol)) {
554 if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
555 ntfs_warning(sb, "Failed to clear dirty bit "
556 "in volume information "
557 "flags. Run chkdsk.");
562 // TODO: Deal with *flags.
564 if (!parse_options(vol, opt))
572 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
573 * @sb: Super block of the device to which @b belongs.
574 * @b: Boot sector of device @sb to check.
575 * @silent: If 'true', all output will be silenced.
577 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
578 * sector. Returns 'true' if it is valid and 'false' if not.
580 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
583 static bool is_boot_sector_ntfs(const struct super_block *sb,
584 const NTFS_BOOT_SECTOR *b, const bool silent)
587 * Check that checksum == sum of u32 values from b to the checksum
588 * field. If checksum is zero, no checking is done. We will work when
589 * the checksum test fails, since some utilities update the boot sector
590 * ignoring the checksum which leaves the checksum out-of-date. We
591 * report a warning if this is the case.
593 if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
597 for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
598 i += le32_to_cpup(u);
599 if (le32_to_cpu(b->checksum) != i)
600 ntfs_warning(sb, "Invalid boot sector checksum.");
602 /* Check OEMidentifier is "NTFS " */
603 if (b->oem_id != magicNTFS)
605 /* Check bytes per sector value is between 256 and 4096. */
606 if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
607 le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
609 /* Check sectors per cluster value is valid. */
610 switch (b->bpb.sectors_per_cluster) {
611 case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
616 /* Check the cluster size is not above the maximum (64kiB). */
617 if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
618 b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
620 /* Check reserved/unused fields are really zero. */
621 if (le16_to_cpu(b->bpb.reserved_sectors) ||
622 le16_to_cpu(b->bpb.root_entries) ||
623 le16_to_cpu(b->bpb.sectors) ||
624 le16_to_cpu(b->bpb.sectors_per_fat) ||
625 le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
627 /* Check clusters per file mft record value is valid. */
628 if ((u8)b->clusters_per_mft_record < 0xe1 ||
629 (u8)b->clusters_per_mft_record > 0xf7)
630 switch (b->clusters_per_mft_record) {
631 case 1: case 2: case 4: case 8: case 16: case 32: case 64:
636 /* Check clusters per index block value is valid. */
637 if ((u8)b->clusters_per_index_record < 0xe1 ||
638 (u8)b->clusters_per_index_record > 0xf7)
639 switch (b->clusters_per_index_record) {
640 case 1: case 2: case 4: case 8: case 16: case 32: case 64:
646 * Check for valid end of sector marker. We will work without it, but
647 * many BIOSes will refuse to boot from a bootsector if the magic is
648 * incorrect, so we emit a warning.
650 if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
651 ntfs_warning(sb, "Invalid end of sector marker.");
658 * read_ntfs_boot_sector - read the NTFS boot sector of a device
659 * @sb: super block of device to read the boot sector from
660 * @silent: if true, suppress all output
662 * Reads the boot sector from the device and validates it. If that fails, tries
663 * to read the backup boot sector, first from the end of the device a-la NT4 and
664 * later and then from the middle of the device a-la NT3.51 and before.
666 * If a valid boot sector is found but it is not the primary boot sector, we
667 * repair the primary boot sector silently (unless the device is read-only or
668 * the primary boot sector is not accessible).
670 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
671 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
672 * to their respective values.
674 * Return the unlocked buffer head containing the boot sector or NULL on error.
676 static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
679 const char *read_err_str = "Unable to read %s boot sector.";
680 struct buffer_head *bh_primary, *bh_backup;
681 sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
683 /* Try to read primary boot sector. */
684 if ((bh_primary = sb_bread(sb, 0))) {
685 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
686 bh_primary->b_data, silent))
689 ntfs_error(sb, "Primary boot sector is invalid.");
691 ntfs_error(sb, read_err_str, "primary");
692 if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
696 ntfs_error(sb, "Mount option errors=recover not used. "
697 "Aborting without trying to recover.");
700 /* Try to read NT4+ backup boot sector. */
701 if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
702 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
703 bh_backup->b_data, silent))
704 goto hotfix_primary_boot_sector;
707 ntfs_error(sb, read_err_str, "backup");
708 /* Try to read NT3.51- backup boot sector. */
709 if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
710 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
711 bh_backup->b_data, silent))
712 goto hotfix_primary_boot_sector;
714 ntfs_error(sb, "Could not find a valid backup boot "
718 ntfs_error(sb, read_err_str, "backup");
719 /* We failed. Cleanup and return. */
723 hotfix_primary_boot_sector:
726 * If we managed to read sector zero and the volume is not
727 * read-only, copy the found, valid backup boot sector to the
728 * primary boot sector. Note we only copy the actual boot
729 * sector structure, not the actual whole device sector as that
730 * may be bigger and would potentially damage the $Boot system
731 * file (FIXME: Would be nice to know if the backup boot sector
732 * on a large sector device contains the whole boot loader or
733 * just the first 512 bytes).
735 if (!sb_rdonly(sb)) {
736 ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
737 "boot sector from backup copy.");
738 memcpy(bh_primary->b_data, bh_backup->b_data,
740 mark_buffer_dirty(bh_primary);
741 sync_dirty_buffer(bh_primary);
742 if (buffer_uptodate(bh_primary)) {
746 ntfs_error(sb, "Hot-fix: Device write error while "
747 "recovering primary boot sector.");
749 ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
750 "sector failed: Read-only mount.");
754 ntfs_warning(sb, "Using backup boot sector.");
759 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
760 * @vol: volume structure to initialise with data from boot sector
761 * @b: boot sector to parse
763 * Parse the ntfs boot sector @b and store all imporant information therein in
764 * the ntfs super block @vol. Return 'true' on success and 'false' on error.
766 static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
768 unsigned int sectors_per_cluster_bits, nr_hidden_sects;
769 int clusters_per_mft_record, clusters_per_index_record;
772 vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
773 vol->sector_size_bits = ffs(vol->sector_size) - 1;
774 ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
776 ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
777 vol->sector_size_bits);
778 if (vol->sector_size < vol->sb->s_blocksize) {
779 ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
780 "device block size (%lu). This is not "
781 "supported. Sorry.", vol->sector_size,
782 vol->sb->s_blocksize);
785 ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
786 sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
787 ntfs_debug("sectors_per_cluster_bits = 0x%x",
788 sectors_per_cluster_bits);
789 nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
790 ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
791 vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
792 vol->cluster_size_mask = vol->cluster_size - 1;
793 vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
794 ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
796 ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
797 ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
798 if (vol->cluster_size < vol->sector_size) {
799 ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
800 "sector size (%i). This is not supported. "
801 "Sorry.", vol->cluster_size, vol->sector_size);
804 clusters_per_mft_record = b->clusters_per_mft_record;
805 ntfs_debug("clusters_per_mft_record = %i (0x%x)",
806 clusters_per_mft_record, clusters_per_mft_record);
807 if (clusters_per_mft_record > 0)
808 vol->mft_record_size = vol->cluster_size <<
809 (ffs(clusters_per_mft_record) - 1);
812 * When mft_record_size < cluster_size, clusters_per_mft_record
813 * = -log2(mft_record_size) bytes. mft_record_size normaly is
814 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
816 vol->mft_record_size = 1 << -clusters_per_mft_record;
817 vol->mft_record_size_mask = vol->mft_record_size - 1;
818 vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
819 ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
820 vol->mft_record_size);
821 ntfs_debug("vol->mft_record_size_mask = 0x%x",
822 vol->mft_record_size_mask);
823 ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
824 vol->mft_record_size_bits, vol->mft_record_size_bits);
826 * We cannot support mft record sizes above the PAGE_SIZE since
827 * we store $MFT/$DATA, the table of mft records in the page cache.
829 if (vol->mft_record_size > PAGE_SIZE) {
830 ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
831 "PAGE_SIZE on your system (%lu). "
832 "This is not supported. Sorry.",
833 vol->mft_record_size, PAGE_SIZE);
836 /* We cannot support mft record sizes below the sector size. */
837 if (vol->mft_record_size < vol->sector_size) {
838 ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
839 "sector size (%i). This is not supported. "
840 "Sorry.", vol->mft_record_size,
844 clusters_per_index_record = b->clusters_per_index_record;
845 ntfs_debug("clusters_per_index_record = %i (0x%x)",
846 clusters_per_index_record, clusters_per_index_record);
847 if (clusters_per_index_record > 0)
848 vol->index_record_size = vol->cluster_size <<
849 (ffs(clusters_per_index_record) - 1);
852 * When index_record_size < cluster_size,
853 * clusters_per_index_record = -log2(index_record_size) bytes.
854 * index_record_size normaly equals 4096 bytes, which is
855 * encoded as 0xF4 (-12 in decimal).
857 vol->index_record_size = 1 << -clusters_per_index_record;
858 vol->index_record_size_mask = vol->index_record_size - 1;
859 vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
860 ntfs_debug("vol->index_record_size = %i (0x%x)",
861 vol->index_record_size, vol->index_record_size);
862 ntfs_debug("vol->index_record_size_mask = 0x%x",
863 vol->index_record_size_mask);
864 ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
865 vol->index_record_size_bits,
866 vol->index_record_size_bits);
867 /* We cannot support index record sizes below the sector size. */
868 if (vol->index_record_size < vol->sector_size) {
869 ntfs_error(vol->sb, "Index record size (%i) is smaller than "
870 "the sector size (%i). This is not "
871 "supported. Sorry.", vol->index_record_size,
876 * Get the size of the volume in clusters and check for 64-bit-ness.
877 * Windows currently only uses 32 bits to save the clusters so we do
878 * the same as it is much faster on 32-bit CPUs.
880 ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
881 if ((u64)ll >= 1ULL << 32) {
882 ntfs_error(vol->sb, "Cannot handle 64-bit clusters. Sorry.");
885 vol->nr_clusters = ll;
886 ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
888 * On an architecture where unsigned long is 32-bits, we restrict the
889 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
890 * will hopefully optimize the whole check away.
892 if (sizeof(unsigned long) < 8) {
893 if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
894 ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
895 "large for this architecture. "
896 "Maximum supported is 2TiB. Sorry.",
897 (unsigned long long)ll >> (40 -
898 vol->cluster_size_bits));
902 ll = sle64_to_cpu(b->mft_lcn);
903 if (ll >= vol->nr_clusters) {
904 ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
905 "volume. Weird.", (unsigned long long)ll,
906 (unsigned long long)ll);
910 ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
911 ll = sle64_to_cpu(b->mftmirr_lcn);
912 if (ll >= vol->nr_clusters) {
913 ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
914 "of volume. Weird.", (unsigned long long)ll,
915 (unsigned long long)ll);
918 vol->mftmirr_lcn = ll;
919 ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
922 * Work out the size of the mft mirror in number of mft records. If the
923 * cluster size is less than or equal to the size taken by four mft
924 * records, the mft mirror stores the first four mft records. If the
925 * cluster size is bigger than the size taken by four mft records, the
926 * mft mirror contains as many mft records as will fit into one
929 if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
930 vol->mftmirr_size = 4;
932 vol->mftmirr_size = vol->cluster_size >>
933 vol->mft_record_size_bits;
934 ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
936 vol->serial_no = le64_to_cpu(b->volume_serial_number);
937 ntfs_debug("vol->serial_no = 0x%llx",
938 (unsigned long long)vol->serial_no);
943 * ntfs_setup_allocators - initialize the cluster and mft allocators
944 * @vol: volume structure for which to setup the allocators
946 * Setup the cluster (lcn) and mft allocators to the starting values.
948 static void ntfs_setup_allocators(ntfs_volume *vol)
951 LCN mft_zone_size, mft_lcn;
954 ntfs_debug("vol->mft_zone_multiplier = 0x%x",
955 vol->mft_zone_multiplier);
957 /* Determine the size of the MFT zone. */
958 mft_zone_size = vol->nr_clusters;
959 switch (vol->mft_zone_multiplier) { /* % of volume size in clusters */
961 mft_zone_size >>= 1; /* 50% */
964 mft_zone_size = (mft_zone_size +
965 (mft_zone_size >> 1)) >> 2; /* 37.5% */
968 mft_zone_size >>= 2; /* 25% */
972 mft_zone_size >>= 3; /* 12.5% */
975 /* Setup the mft zone. */
976 vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
977 ntfs_debug("vol->mft_zone_pos = 0x%llx",
978 (unsigned long long)vol->mft_zone_pos);
980 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
981 * source) and if the actual mft_lcn is in the expected place or even
982 * further to the front of the volume, extend the mft_zone to cover the
983 * beginning of the volume as well. This is in order to protect the
984 * area reserved for the mft bitmap as well within the mft_zone itself.
985 * On non-standard volumes we do not protect it as the overhead would
986 * be higher than the speed increase we would get by doing it.
988 mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
989 if (mft_lcn * vol->cluster_size < 16 * 1024)
990 mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
992 if (vol->mft_zone_start <= mft_lcn)
993 vol->mft_zone_start = 0;
994 ntfs_debug("vol->mft_zone_start = 0x%llx",
995 (unsigned long long)vol->mft_zone_start);
997 * Need to cap the mft zone on non-standard volumes so that it does
998 * not point outside the boundaries of the volume. We do this by
999 * halving the zone size until we are inside the volume.
1001 vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
1002 while (vol->mft_zone_end >= vol->nr_clusters) {
1003 mft_zone_size >>= 1;
1004 vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
1006 ntfs_debug("vol->mft_zone_end = 0x%llx",
1007 (unsigned long long)vol->mft_zone_end);
1009 * Set the current position within each data zone to the start of the
1012 vol->data1_zone_pos = vol->mft_zone_end;
1013 ntfs_debug("vol->data1_zone_pos = 0x%llx",
1014 (unsigned long long)vol->data1_zone_pos);
1015 vol->data2_zone_pos = 0;
1016 ntfs_debug("vol->data2_zone_pos = 0x%llx",
1017 (unsigned long long)vol->data2_zone_pos);
1019 /* Set the mft data allocation position to mft record 24. */
1020 vol->mft_data_pos = 24;
1021 ntfs_debug("vol->mft_data_pos = 0x%llx",
1022 (unsigned long long)vol->mft_data_pos);
1023 #endif /* NTFS_RW */
1029 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
1030 * @vol: ntfs super block describing device whose mft mirror to load
1032 * Return 'true' on success or 'false' on error.
1034 static bool load_and_init_mft_mirror(ntfs_volume *vol)
1036 struct inode *tmp_ino;
1039 ntfs_debug("Entering.");
1040 /* Get mft mirror inode. */
1041 tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
1042 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1043 if (!IS_ERR(tmp_ino))
1045 /* Caller will display error message. */
1049 * Re-initialize some specifics about $MFTMirr's inode as
1050 * ntfs_read_inode() will have set up the default ones.
1052 /* Set uid and gid to root. */
1053 tmp_ino->i_uid = GLOBAL_ROOT_UID;
1054 tmp_ino->i_gid = GLOBAL_ROOT_GID;
1055 /* Regular file. No access for anyone. */
1056 tmp_ino->i_mode = S_IFREG;
1057 /* No VFS initiated operations allowed for $MFTMirr. */
1058 tmp_ino->i_op = &ntfs_empty_inode_ops;
1059 tmp_ino->i_fop = &ntfs_empty_file_ops;
1060 /* Put in our special address space operations. */
1061 tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1062 tmp_ni = NTFS_I(tmp_ino);
1063 /* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1064 NInoSetMstProtected(tmp_ni);
1065 NInoSetSparseDisabled(tmp_ni);
1067 * Set up our little cheat allowing us to reuse the async read io
1068 * completion handler for directories.
1070 tmp_ni->itype.index.block_size = vol->mft_record_size;
1071 tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1072 vol->mftmirr_ino = tmp_ino;
1073 ntfs_debug("Done.");
1078 * check_mft_mirror - compare contents of the mft mirror with the mft
1079 * @vol: ntfs super block describing device whose mft mirror to check
1081 * Return 'true' on success or 'false' on error.
1083 * Note, this function also results in the mft mirror runlist being completely
1084 * mapped into memory. The mft mirror write code requires this and will BUG()
1085 * should it find an unmapped runlist element.
1087 static bool check_mft_mirror(ntfs_volume *vol)
1089 struct super_block *sb = vol->sb;
1090 ntfs_inode *mirr_ni;
1091 struct page *mft_page, *mirr_page;
1093 runlist_element *rl, rl2[2];
1095 int mrecs_per_page, i;
1097 ntfs_debug("Entering.");
1098 /* Compare contents of $MFT and $MFTMirr. */
1099 mrecs_per_page = PAGE_SIZE / vol->mft_record_size;
1100 BUG_ON(!mrecs_per_page);
1101 BUG_ON(!vol->mftmirr_size);
1102 mft_page = mirr_page = NULL;
1103 kmft = kmirr = NULL;
1108 /* Switch pages if necessary. */
1109 if (!(i % mrecs_per_page)) {
1111 ntfs_unmap_page(mft_page);
1112 ntfs_unmap_page(mirr_page);
1114 /* Get the $MFT page. */
1115 mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1117 if (IS_ERR(mft_page)) {
1118 ntfs_error(sb, "Failed to read $MFT.");
1121 kmft = page_address(mft_page);
1122 /* Get the $MFTMirr page. */
1123 mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1125 if (IS_ERR(mirr_page)) {
1126 ntfs_error(sb, "Failed to read $MFTMirr.");
1129 kmirr = page_address(mirr_page);
1132 /* Do not check the record if it is not in use. */
1133 if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1134 /* Make sure the record is ok. */
1135 if (ntfs_is_baad_recordp((le32*)kmft)) {
1136 ntfs_error(sb, "Incomplete multi sector "
1137 "transfer detected in mft "
1140 ntfs_unmap_page(mirr_page);
1142 ntfs_unmap_page(mft_page);
1146 /* Do not check the mirror record if it is not in use. */
1147 if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1148 if (ntfs_is_baad_recordp((le32*)kmirr)) {
1149 ntfs_error(sb, "Incomplete multi sector "
1150 "transfer detected in mft "
1151 "mirror record %i.", i);
1155 /* Get the amount of data in the current record. */
1156 bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1157 if (bytes < sizeof(MFT_RECORD_OLD) ||
1158 bytes > vol->mft_record_size ||
1159 ntfs_is_baad_recordp((le32*)kmft)) {
1160 bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1161 if (bytes < sizeof(MFT_RECORD_OLD) ||
1162 bytes > vol->mft_record_size ||
1163 ntfs_is_baad_recordp((le32*)kmirr))
1164 bytes = vol->mft_record_size;
1166 /* Compare the two records. */
1167 if (memcmp(kmft, kmirr, bytes)) {
1168 ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1169 "match. Run ntfsfix or chkdsk.", i);
1172 kmft += vol->mft_record_size;
1173 kmirr += vol->mft_record_size;
1174 } while (++i < vol->mftmirr_size);
1175 /* Release the last pages. */
1176 ntfs_unmap_page(mft_page);
1177 ntfs_unmap_page(mirr_page);
1179 /* Construct the mft mirror runlist by hand. */
1181 rl2[0].lcn = vol->mftmirr_lcn;
1182 rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1183 vol->cluster_size - 1) / vol->cluster_size;
1184 rl2[1].vcn = rl2[0].length;
1185 rl2[1].lcn = LCN_ENOENT;
1188 * Because we have just read all of the mft mirror, we know we have
1189 * mapped the full runlist for it.
1191 mirr_ni = NTFS_I(vol->mftmirr_ino);
1192 down_read(&mirr_ni->runlist.lock);
1193 rl = mirr_ni->runlist.rl;
1194 /* Compare the two runlists. They must be identical. */
1197 if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1198 rl2[i].length != rl[i].length) {
1199 ntfs_error(sb, "$MFTMirr location mismatch. "
1201 up_read(&mirr_ni->runlist.lock);
1204 } while (rl2[i++].length);
1205 up_read(&mirr_ni->runlist.lock);
1206 ntfs_debug("Done.");
1211 * load_and_check_logfile - load and check the logfile inode for a volume
1212 * @vol: ntfs super block describing device whose logfile to load
1214 * Return 'true' on success or 'false' on error.
1216 static bool load_and_check_logfile(ntfs_volume *vol,
1217 RESTART_PAGE_HEADER **rp)
1219 struct inode *tmp_ino;
1221 ntfs_debug("Entering.");
1222 tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1223 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1224 if (!IS_ERR(tmp_ino))
1226 /* Caller will display error message. */
1229 if (!ntfs_check_logfile(tmp_ino, rp)) {
1231 /* ntfs_check_logfile() will have displayed error output. */
1234 NInoSetSparseDisabled(NTFS_I(tmp_ino));
1235 vol->logfile_ino = tmp_ino;
1236 ntfs_debug("Done.");
1240 #define NTFS_HIBERFIL_HEADER_SIZE 4096
1243 * check_windows_hibernation_status - check if Windows is suspended on a volume
1244 * @vol: ntfs super block of device to check
1246 * Check if Windows is hibernated on the ntfs volume @vol. This is done by
1247 * looking for the file hiberfil.sys in the root directory of the volume. If
1248 * the file is not present Windows is definitely not suspended.
1250 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1251 * definitely suspended (this volume is not the system volume). Caveat: on a
1252 * system with many volumes it is possible that the < 4kiB check is bogus but
1253 * for now this should do fine.
1255 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1256 * hiberfil header (which is the first 4kiB). If this begins with "hibr",
1257 * Windows is definitely suspended. If it is completely full of zeroes,
1258 * Windows is definitely not hibernated. Any other case is treated as if
1259 * Windows is suspended. This caters for the above mentioned caveat of a
1260 * system with many volumes where no "hibr" magic would be present and there is
1263 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1264 * hibernated on the volume, and -errno on error.
1266 static int check_windows_hibernation_status(ntfs_volume *vol)
1272 ntfs_name *name = NULL;
1274 static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
1275 cpu_to_le16('i'), cpu_to_le16('b'),
1276 cpu_to_le16('e'), cpu_to_le16('r'),
1277 cpu_to_le16('f'), cpu_to_le16('i'),
1278 cpu_to_le16('l'), cpu_to_le16('.'),
1279 cpu_to_le16('s'), cpu_to_le16('y'),
1280 cpu_to_le16('s'), 0 };
1282 ntfs_debug("Entering.");
1284 * Find the inode number for the hibernation file by looking up the
1285 * filename hiberfil.sys in the root directory.
1287 inode_lock(vol->root_ino);
1288 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1290 inode_unlock(vol->root_ino);
1291 if (IS_ERR_MREF(mref)) {
1292 ret = MREF_ERR(mref);
1293 /* If the file does not exist, Windows is not hibernated. */
1294 if (ret == -ENOENT) {
1295 ntfs_debug("hiberfil.sys not present. Windows is not "
1296 "hibernated on the volume.");
1299 /* A real error occurred. */
1300 ntfs_error(vol->sb, "Failed to find inode number for "
1304 /* We do not care for the type of match that was found. */
1306 /* Get the inode. */
1307 vi = ntfs_iget(vol->sb, MREF(mref));
1308 if (IS_ERR(vi) || is_bad_inode(vi)) {
1311 ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1312 return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1314 if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1315 ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx). "
1316 "Windows is hibernated on the volume. This "
1317 "is not the system volume.", i_size_read(vi));
1320 page = ntfs_map_page(vi->i_mapping, 0);
1322 ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1323 ret = PTR_ERR(page);
1326 kaddr = (u32*)page_address(page);
1327 if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
1328 ntfs_debug("Magic \"hibr\" found in hiberfil.sys. Windows is "
1329 "hibernated on the volume. This is the "
1333 kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1335 if (unlikely(*kaddr)) {
1336 ntfs_debug("hiberfil.sys is larger than 4kiB "
1337 "(0x%llx), does not contain the "
1338 "\"hibr\" magic, and does not have a "
1339 "zero header. Windows is hibernated "
1340 "on the volume. This is not the "
1341 "system volume.", i_size_read(vi));
1344 } while (++kaddr < kend);
1345 ntfs_debug("hiberfil.sys contains a zero header. Windows is not "
1346 "hibernated on the volume. This is the system "
1350 ntfs_unmap_page(page);
1357 * load_and_init_quota - load and setup the quota file for a volume if present
1358 * @vol: ntfs super block describing device whose quota file to load
1360 * Return 'true' on success or 'false' on error. If $Quota is not present, we
1361 * leave vol->quota_ino as NULL and return success.
1363 static bool load_and_init_quota(ntfs_volume *vol)
1366 struct inode *tmp_ino;
1367 ntfs_name *name = NULL;
1368 static const ntfschar Quota[7] = { cpu_to_le16('$'),
1369 cpu_to_le16('Q'), cpu_to_le16('u'),
1370 cpu_to_le16('o'), cpu_to_le16('t'),
1371 cpu_to_le16('a'), 0 };
1372 static ntfschar Q[3] = { cpu_to_le16('$'),
1373 cpu_to_le16('Q'), 0 };
1375 ntfs_debug("Entering.");
1377 * Find the inode number for the quota file by looking up the filename
1378 * $Quota in the extended system files directory $Extend.
1380 inode_lock(vol->extend_ino);
1381 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1383 inode_unlock(vol->extend_ino);
1384 if (IS_ERR_MREF(mref)) {
1386 * If the file does not exist, quotas are disabled and have
1387 * never been enabled on this volume, just return success.
1389 if (MREF_ERR(mref) == -ENOENT) {
1390 ntfs_debug("$Quota not present. Volume does not have "
1393 * No need to try to set quotas out of date if they are
1396 NVolSetQuotaOutOfDate(vol);
1399 /* A real error occurred. */
1400 ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1403 /* We do not care for the type of match that was found. */
1405 /* Get the inode. */
1406 tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1407 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1408 if (!IS_ERR(tmp_ino))
1410 ntfs_error(vol->sb, "Failed to load $Quota.");
1413 vol->quota_ino = tmp_ino;
1414 /* Get the $Q index allocation attribute. */
1415 tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1416 if (IS_ERR(tmp_ino)) {
1417 ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1420 vol->quota_q_ino = tmp_ino;
1421 ntfs_debug("Done.");
1426 * load_and_init_usnjrnl - load and setup the transaction log if present
1427 * @vol: ntfs super block describing device whose usnjrnl file to load
1429 * Return 'true' on success or 'false' on error.
1431 * If $UsnJrnl is not present or in the process of being disabled, we set
1432 * NVolUsnJrnlStamped() and return success.
1434 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1435 * i.e. transaction logging has only just been enabled or the journal has been
1436 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1437 * and return success.
1439 static bool load_and_init_usnjrnl(ntfs_volume *vol)
1442 struct inode *tmp_ino;
1445 ntfs_name *name = NULL;
1447 static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
1448 cpu_to_le16('U'), cpu_to_le16('s'),
1449 cpu_to_le16('n'), cpu_to_le16('J'),
1450 cpu_to_le16('r'), cpu_to_le16('n'),
1451 cpu_to_le16('l'), 0 };
1452 static ntfschar Max[5] = { cpu_to_le16('$'),
1453 cpu_to_le16('M'), cpu_to_le16('a'),
1454 cpu_to_le16('x'), 0 };
1455 static ntfschar J[3] = { cpu_to_le16('$'),
1456 cpu_to_le16('J'), 0 };
1458 ntfs_debug("Entering.");
1460 * Find the inode number for the transaction log file by looking up the
1461 * filename $UsnJrnl in the extended system files directory $Extend.
1463 inode_lock(vol->extend_ino);
1464 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1466 inode_unlock(vol->extend_ino);
1467 if (IS_ERR_MREF(mref)) {
1469 * If the file does not exist, transaction logging is disabled,
1470 * just return success.
1472 if (MREF_ERR(mref) == -ENOENT) {
1473 ntfs_debug("$UsnJrnl not present. Volume does not "
1474 "have transaction logging enabled.");
1477 * No need to try to stamp the transaction log if
1478 * transaction logging is not enabled.
1480 NVolSetUsnJrnlStamped(vol);
1483 /* A real error occurred. */
1484 ntfs_error(vol->sb, "Failed to find inode number for "
1488 /* We do not care for the type of match that was found. */
1490 /* Get the inode. */
1491 tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1492 if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1493 if (!IS_ERR(tmp_ino))
1495 ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1498 vol->usnjrnl_ino = tmp_ino;
1500 * If the transaction log is in the process of being deleted, we can
1503 if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1504 ntfs_debug("$UsnJrnl in the process of being disabled. "
1505 "Volume does not have transaction logging "
1509 /* Get the $DATA/$Max attribute. */
1510 tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1511 if (IS_ERR(tmp_ino)) {
1512 ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1516 vol->usnjrnl_max_ino = tmp_ino;
1517 if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1518 ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1519 "attribute (size is 0x%llx but should be at "
1520 "least 0x%zx bytes).", i_size_read(tmp_ino),
1521 sizeof(USN_HEADER));
1524 /* Get the $DATA/$J attribute. */
1525 tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1526 if (IS_ERR(tmp_ino)) {
1527 ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1531 vol->usnjrnl_j_ino = tmp_ino;
1532 /* Verify $J is non-resident and sparse. */
1533 tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1534 if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1535 ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1536 "and/or not sparse.");
1539 /* Read the USN_HEADER from $DATA/$Max. */
1540 page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1542 ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1546 uh = (USN_HEADER*)page_address(page);
1547 /* Sanity check the $Max. */
1548 if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1549 sle64_to_cpu(uh->maximum_size))) {
1550 ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1551 "maximum size (0x%llx). $UsnJrnl is corrupt.",
1552 (long long)sle64_to_cpu(uh->allocation_delta),
1553 (long long)sle64_to_cpu(uh->maximum_size));
1554 ntfs_unmap_page(page);
1558 * If the transaction log has been stamped and nothing has been written
1559 * to it since, we do not need to stamp it.
1561 if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1562 i_size_read(vol->usnjrnl_j_ino))) {
1563 if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1564 i_size_read(vol->usnjrnl_j_ino))) {
1565 ntfs_unmap_page(page);
1566 ntfs_debug("$UsnJrnl is enabled but nothing has been "
1567 "logged since it was last stamped. "
1568 "Treating this as if the volume does "
1569 "not have transaction logging "
1573 ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1574 "which is out of bounds (0x%llx). $UsnJrnl "
1576 (long long)sle64_to_cpu(uh->lowest_valid_usn),
1577 i_size_read(vol->usnjrnl_j_ino));
1578 ntfs_unmap_page(page);
1581 ntfs_unmap_page(page);
1582 ntfs_debug("Done.");
1587 * load_and_init_attrdef - load the attribute definitions table for a volume
1588 * @vol: ntfs super block describing device whose attrdef to load
1590 * Return 'true' on success or 'false' on error.
1592 static bool load_and_init_attrdef(ntfs_volume *vol)
1595 struct super_block *sb = vol->sb;
1598 pgoff_t index, max_index;
1601 ntfs_debug("Entering.");
1602 /* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1603 ino = ntfs_iget(sb, FILE_AttrDef);
1604 if (IS_ERR(ino) || is_bad_inode(ino)) {
1609 NInoSetSparseDisabled(NTFS_I(ino));
1610 /* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1611 i_size = i_size_read(ino);
1612 if (i_size <= 0 || i_size > 0x7fffffff)
1614 vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1618 max_index = i_size >> PAGE_SHIFT;
1620 while (index < max_index) {
1621 /* Read the attrdef table and copy it into the linear buffer. */
1622 read_partial_attrdef_page:
1623 page = ntfs_map_page(ino->i_mapping, index);
1625 goto free_iput_failed;
1626 memcpy((u8*)vol->attrdef + (index++ << PAGE_SHIFT),
1627 page_address(page), size);
1628 ntfs_unmap_page(page);
1630 if (size == PAGE_SIZE) {
1631 size = i_size & ~PAGE_MASK;
1633 goto read_partial_attrdef_page;
1635 vol->attrdef_size = i_size;
1636 ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1640 ntfs_free(vol->attrdef);
1641 vol->attrdef = NULL;
1645 ntfs_error(sb, "Failed to initialize attribute definition table.");
1649 #endif /* NTFS_RW */
1652 * load_and_init_upcase - load the upcase table for an ntfs volume
1653 * @vol: ntfs super block describing device whose upcase to load
1655 * Return 'true' on success or 'false' on error.
1657 static bool load_and_init_upcase(ntfs_volume *vol)
1660 struct super_block *sb = vol->sb;
1663 pgoff_t index, max_index;
1667 ntfs_debug("Entering.");
1668 /* Read upcase table and setup vol->upcase and vol->upcase_len. */
1669 ino = ntfs_iget(sb, FILE_UpCase);
1670 if (IS_ERR(ino) || is_bad_inode(ino)) {
1676 * The upcase size must not be above 64k Unicode characters, must not
1677 * be zero and must be a multiple of sizeof(ntfschar).
1679 i_size = i_size_read(ino);
1680 if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1681 i_size > 64ULL * 1024 * sizeof(ntfschar))
1682 goto iput_upcase_failed;
1683 vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1685 goto iput_upcase_failed;
1687 max_index = i_size >> PAGE_SHIFT;
1689 while (index < max_index) {
1690 /* Read the upcase table and copy it into the linear buffer. */
1691 read_partial_upcase_page:
1692 page = ntfs_map_page(ino->i_mapping, index);
1694 goto iput_upcase_failed;
1695 memcpy((char*)vol->upcase + (index++ << PAGE_SHIFT),
1696 page_address(page), size);
1697 ntfs_unmap_page(page);
1699 if (size == PAGE_SIZE) {
1700 size = i_size & ~PAGE_MASK;
1702 goto read_partial_upcase_page;
1704 vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1705 ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1706 i_size, 64 * 1024 * sizeof(ntfschar));
1708 mutex_lock(&ntfs_lock);
1709 if (!default_upcase) {
1710 ntfs_debug("Using volume specified $UpCase since default is "
1712 mutex_unlock(&ntfs_lock);
1715 max = default_upcase_len;
1716 if (max > vol->upcase_len)
1717 max = vol->upcase_len;
1718 for (i = 0; i < max; i++)
1719 if (vol->upcase[i] != default_upcase[i])
1722 ntfs_free(vol->upcase);
1723 vol->upcase = default_upcase;
1724 vol->upcase_len = max;
1725 ntfs_nr_upcase_users++;
1726 mutex_unlock(&ntfs_lock);
1727 ntfs_debug("Volume specified $UpCase matches default. Using "
1731 mutex_unlock(&ntfs_lock);
1732 ntfs_debug("Using volume specified $UpCase since it does not match "
1737 ntfs_free(vol->upcase);
1740 mutex_lock(&ntfs_lock);
1741 if (default_upcase) {
1742 vol->upcase = default_upcase;
1743 vol->upcase_len = default_upcase_len;
1744 ntfs_nr_upcase_users++;
1745 mutex_unlock(&ntfs_lock);
1746 ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1750 mutex_unlock(&ntfs_lock);
1751 ntfs_error(sb, "Failed to initialize upcase table.");
1756 * The lcn and mft bitmap inodes are NTFS-internal inodes with
1757 * their own special locking rules:
1759 static struct lock_class_key
1760 lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1761 mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1764 * load_system_files - open the system files using normal functions
1765 * @vol: ntfs super block describing device whose system files to load
1767 * Open the system files with normal access functions and complete setting up
1768 * the ntfs super block @vol.
1770 * Return 'true' on success or 'false' on error.
1772 static bool load_system_files(ntfs_volume *vol)
1774 struct super_block *sb = vol->sb;
1776 VOLUME_INFORMATION *vi;
1777 ntfs_attr_search_ctx *ctx;
1779 RESTART_PAGE_HEADER *rp;
1781 #endif /* NTFS_RW */
1783 ntfs_debug("Entering.");
1785 /* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1786 if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1787 static const char *es1 = "Failed to load $MFTMirr";
1788 static const char *es2 = "$MFTMirr does not match $MFT";
1789 static const char *es3 = ". Run ntfsfix and/or chkdsk.";
1791 /* If a read-write mount, convert it to a read-only mount. */
1792 if (!sb_rdonly(sb)) {
1793 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1794 ON_ERRORS_CONTINUE))) {
1795 ntfs_error(sb, "%s and neither on_errors="
1796 "continue nor on_errors="
1797 "remount-ro was specified%s",
1798 !vol->mftmirr_ino ? es1 : es2,
1800 goto iput_mirr_err_out;
1802 sb->s_flags |= SB_RDONLY;
1803 ntfs_error(sb, "%s. Mounting read-only%s",
1804 !vol->mftmirr_ino ? es1 : es2, es3);
1806 ntfs_warning(sb, "%s. Will not be able to remount "
1808 !vol->mftmirr_ino ? es1 : es2, es3);
1809 /* This will prevent a read-write remount. */
1812 #endif /* NTFS_RW */
1813 /* Get mft bitmap attribute inode. */
1814 vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1815 if (IS_ERR(vol->mftbmp_ino)) {
1816 ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1817 goto iput_mirr_err_out;
1819 lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1820 &mftbmp_runlist_lock_key);
1821 lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1822 &mftbmp_mrec_lock_key);
1823 /* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1824 if (!load_and_init_upcase(vol))
1825 goto iput_mftbmp_err_out;
1828 * Read attribute definitions table and setup @vol->attrdef and
1829 * @vol->attrdef_size.
1831 if (!load_and_init_attrdef(vol))
1832 goto iput_upcase_err_out;
1833 #endif /* NTFS_RW */
1835 * Get the cluster allocation bitmap inode and verify the size, no
1836 * need for any locking at this stage as we are already running
1837 * exclusively as we are mount in progress task.
1839 vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1840 if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1841 if (!IS_ERR(vol->lcnbmp_ino))
1842 iput(vol->lcnbmp_ino);
1845 lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1846 &lcnbmp_runlist_lock_key);
1847 lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1848 &lcnbmp_mrec_lock_key);
1850 NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1851 if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1852 iput(vol->lcnbmp_ino);
1854 ntfs_error(sb, "Failed to load $Bitmap.");
1855 goto iput_attrdef_err_out;
1858 * Get the volume inode and setup our cache of the volume flags and
1861 vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1862 if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1863 if (!IS_ERR(vol->vol_ino))
1866 ntfs_error(sb, "Failed to load $Volume.");
1867 goto iput_lcnbmp_err_out;
1869 m = map_mft_record(NTFS_I(vol->vol_ino));
1875 if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1876 ntfs_error(sb, "Failed to get attribute search context.");
1877 goto get_ctx_vol_failed;
1879 if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1880 ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1882 ntfs_attr_put_search_ctx(ctx);
1884 unmap_mft_record(NTFS_I(vol->vol_ino));
1885 goto iput_volume_failed;
1887 vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1888 le16_to_cpu(ctx->attr->data.resident.value_offset));
1889 /* Some bounds checks. */
1890 if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1891 le32_to_cpu(ctx->attr->data.resident.value_length) >
1892 (u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1894 /* Copy the volume flags and version to the ntfs_volume structure. */
1895 vol->vol_flags = vi->flags;
1896 vol->major_ver = vi->major_ver;
1897 vol->minor_ver = vi->minor_ver;
1898 ntfs_attr_put_search_ctx(ctx);
1899 unmap_mft_record(NTFS_I(vol->vol_ino));
1900 pr_info("volume version %i.%i.\n", vol->major_ver,
1902 if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1903 ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1904 "volume version %i.%i (need at least version "
1905 "3.0).", vol->major_ver, vol->minor_ver);
1906 NVolClearSparseEnabled(vol);
1909 /* Make sure that no unsupported volume flags are set. */
1910 if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1911 static const char *es1a = "Volume is dirty";
1912 static const char *es1b = "Volume has been modified by chkdsk";
1913 static const char *es1c = "Volume has unsupported flags set";
1914 static const char *es2a = ". Run chkdsk and mount in Windows.";
1915 static const char *es2b = ". Mount in Windows.";
1916 const char *es1, *es2;
1919 if (vol->vol_flags & VOLUME_IS_DIRTY)
1921 else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1926 ntfs_warning(sb, "Unsupported volume flags 0x%x "
1928 (unsigned)le16_to_cpu(vol->vol_flags));
1930 /* If a read-write mount, convert it to a read-only mount. */
1931 if (!sb_rdonly(sb)) {
1932 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1933 ON_ERRORS_CONTINUE))) {
1934 ntfs_error(sb, "%s and neither on_errors="
1935 "continue nor on_errors="
1936 "remount-ro was specified%s",
1938 goto iput_vol_err_out;
1940 sb->s_flags |= SB_RDONLY;
1941 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1943 ntfs_warning(sb, "%s. Will not be able to remount "
1944 "read-write%s", es1, es2);
1946 * Do not set NVolErrors() because ntfs_remount() re-checks the
1947 * flags which we need to do in case any flags have changed.
1951 * Get the inode for the logfile, check it and determine if the volume
1952 * was shutdown cleanly.
1955 if (!load_and_check_logfile(vol, &rp) ||
1956 !ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1957 static const char *es1a = "Failed to load $LogFile";
1958 static const char *es1b = "$LogFile is not clean";
1959 static const char *es2 = ". Mount in Windows.";
1962 es1 = !vol->logfile_ino ? es1a : es1b;
1963 /* If a read-write mount, convert it to a read-only mount. */
1964 if (!sb_rdonly(sb)) {
1965 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1966 ON_ERRORS_CONTINUE))) {
1967 ntfs_error(sb, "%s and neither on_errors="
1968 "continue nor on_errors="
1969 "remount-ro was specified%s",
1971 if (vol->logfile_ino) {
1975 goto iput_logfile_err_out;
1977 sb->s_flags |= SB_RDONLY;
1978 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1980 ntfs_warning(sb, "%s. Will not be able to remount "
1981 "read-write%s", es1, es2);
1982 /* This will prevent a read-write remount. */
1986 #endif /* NTFS_RW */
1987 /* Get the root directory inode so we can do path lookups. */
1988 vol->root_ino = ntfs_iget(sb, FILE_root);
1989 if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1990 if (!IS_ERR(vol->root_ino))
1991 iput(vol->root_ino);
1992 ntfs_error(sb, "Failed to load root directory.");
1993 goto iput_logfile_err_out;
1997 * Check if Windows is suspended to disk on the target volume. If it
1998 * is hibernated, we must not write *anything* to the disk so set
1999 * NVolErrors() without setting the dirty volume flag and mount
2000 * read-only. This will prevent read-write remounting and it will also
2001 * prevent all writes.
2003 err = check_windows_hibernation_status(vol);
2004 if (unlikely(err)) {
2005 static const char *es1a = "Failed to determine if Windows is "
2007 static const char *es1b = "Windows is hibernated";
2008 static const char *es2 = ". Run chkdsk.";
2011 es1 = err < 0 ? es1a : es1b;
2012 /* If a read-write mount, convert it to a read-only mount. */
2013 if (!sb_rdonly(sb)) {
2014 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2015 ON_ERRORS_CONTINUE))) {
2016 ntfs_error(sb, "%s and neither on_errors="
2017 "continue nor on_errors="
2018 "remount-ro was specified%s",
2020 goto iput_root_err_out;
2022 sb->s_flags |= SB_RDONLY;
2023 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2025 ntfs_warning(sb, "%s. Will not be able to remount "
2026 "read-write%s", es1, es2);
2027 /* This will prevent a read-write remount. */
2030 /* If (still) a read-write mount, mark the volume dirty. */
2031 if (!sb_rdonly(sb) && ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
2032 static const char *es1 = "Failed to set dirty bit in volume "
2033 "information flags";
2034 static const char *es2 = ". Run chkdsk.";
2036 /* Convert to a read-only mount. */
2037 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2038 ON_ERRORS_CONTINUE))) {
2039 ntfs_error(sb, "%s and neither on_errors=continue nor "
2040 "on_errors=remount-ro was specified%s",
2042 goto iput_root_err_out;
2044 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2045 sb->s_flags |= SB_RDONLY;
2047 * Do not set NVolErrors() because ntfs_remount() might manage
2048 * to set the dirty flag in which case all would be well.
2052 // TODO: Enable this code once we start modifying anything that is
2053 // different between NTFS 1.2 and 3.x...
2055 * If (still) a read-write mount, set the NT4 compatibility flag on
2056 * newer NTFS version volumes.
2058 if (!(sb->s_flags & SB_RDONLY) && (vol->major_ver > 1) &&
2059 ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
2060 static const char *es1 = "Failed to set NT4 compatibility flag";
2061 static const char *es2 = ". Run chkdsk.";
2063 /* Convert to a read-only mount. */
2064 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2065 ON_ERRORS_CONTINUE))) {
2066 ntfs_error(sb, "%s and neither on_errors=continue nor "
2067 "on_errors=remount-ro was specified%s",
2069 goto iput_root_err_out;
2071 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2072 sb->s_flags |= SB_RDONLY;
2076 /* If (still) a read-write mount, empty the logfile. */
2077 if (!sb_rdonly(sb) && !ntfs_empty_logfile(vol->logfile_ino)) {
2078 static const char *es1 = "Failed to empty $LogFile";
2079 static const char *es2 = ". Mount in Windows.";
2081 /* Convert to a read-only mount. */
2082 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2083 ON_ERRORS_CONTINUE))) {
2084 ntfs_error(sb, "%s and neither on_errors=continue nor "
2085 "on_errors=remount-ro was specified%s",
2087 goto iput_root_err_out;
2089 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2090 sb->s_flags |= SB_RDONLY;
2093 #endif /* NTFS_RW */
2094 /* If on NTFS versions before 3.0, we are done. */
2095 if (unlikely(vol->major_ver < 3))
2097 /* NTFS 3.0+ specific initialization. */
2098 /* Get the security descriptors inode. */
2099 vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2100 if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2101 if (!IS_ERR(vol->secure_ino))
2102 iput(vol->secure_ino);
2103 ntfs_error(sb, "Failed to load $Secure.");
2104 goto iput_root_err_out;
2106 // TODO: Initialize security.
2107 /* Get the extended system files' directory inode. */
2108 vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2109 if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2110 if (!IS_ERR(vol->extend_ino))
2111 iput(vol->extend_ino);
2112 ntfs_error(sb, "Failed to load $Extend.");
2113 goto iput_sec_err_out;
2116 /* Find the quota file, load it if present, and set it up. */
2117 if (!load_and_init_quota(vol)) {
2118 static const char *es1 = "Failed to load $Quota";
2119 static const char *es2 = ". Run chkdsk.";
2121 /* If a read-write mount, convert it to a read-only mount. */
2122 if (!sb_rdonly(sb)) {
2123 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2124 ON_ERRORS_CONTINUE))) {
2125 ntfs_error(sb, "%s and neither on_errors="
2126 "continue nor on_errors="
2127 "remount-ro was specified%s",
2129 goto iput_quota_err_out;
2131 sb->s_flags |= SB_RDONLY;
2132 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2134 ntfs_warning(sb, "%s. Will not be able to remount "
2135 "read-write%s", es1, es2);
2136 /* This will prevent a read-write remount. */
2139 /* If (still) a read-write mount, mark the quotas out of date. */
2140 if (!sb_rdonly(sb) && !ntfs_mark_quotas_out_of_date(vol)) {
2141 static const char *es1 = "Failed to mark quotas out of date";
2142 static const char *es2 = ". Run chkdsk.";
2144 /* Convert to a read-only mount. */
2145 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2146 ON_ERRORS_CONTINUE))) {
2147 ntfs_error(sb, "%s and neither on_errors=continue nor "
2148 "on_errors=remount-ro was specified%s",
2150 goto iput_quota_err_out;
2152 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2153 sb->s_flags |= SB_RDONLY;
2157 * Find the transaction log file ($UsnJrnl), load it if present, check
2158 * it, and set it up.
2160 if (!load_and_init_usnjrnl(vol)) {
2161 static const char *es1 = "Failed to load $UsnJrnl";
2162 static const char *es2 = ". Run chkdsk.";
2164 /* If a read-write mount, convert it to a read-only mount. */
2165 if (!sb_rdonly(sb)) {
2166 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2167 ON_ERRORS_CONTINUE))) {
2168 ntfs_error(sb, "%s and neither on_errors="
2169 "continue nor on_errors="
2170 "remount-ro was specified%s",
2172 goto iput_usnjrnl_err_out;
2174 sb->s_flags |= SB_RDONLY;
2175 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2177 ntfs_warning(sb, "%s. Will not be able to remount "
2178 "read-write%s", es1, es2);
2179 /* This will prevent a read-write remount. */
2182 /* If (still) a read-write mount, stamp the transaction log. */
2183 if (!sb_rdonly(sb) && !ntfs_stamp_usnjrnl(vol)) {
2184 static const char *es1 = "Failed to stamp transaction log "
2186 static const char *es2 = ". Run chkdsk.";
2188 /* Convert to a read-only mount. */
2189 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2190 ON_ERRORS_CONTINUE))) {
2191 ntfs_error(sb, "%s and neither on_errors=continue nor "
2192 "on_errors=remount-ro was specified%s",
2194 goto iput_usnjrnl_err_out;
2196 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2197 sb->s_flags |= SB_RDONLY;
2200 #endif /* NTFS_RW */
2203 iput_usnjrnl_err_out:
2204 iput(vol->usnjrnl_j_ino);
2205 iput(vol->usnjrnl_max_ino);
2206 iput(vol->usnjrnl_ino);
2208 iput(vol->quota_q_ino);
2209 iput(vol->quota_ino);
2210 iput(vol->extend_ino);
2211 #endif /* NTFS_RW */
2213 iput(vol->secure_ino);
2215 iput(vol->root_ino);
2216 iput_logfile_err_out:
2218 iput(vol->logfile_ino);
2220 #endif /* NTFS_RW */
2222 iput_lcnbmp_err_out:
2223 iput(vol->lcnbmp_ino);
2224 iput_attrdef_err_out:
2225 vol->attrdef_size = 0;
2227 ntfs_free(vol->attrdef);
2228 vol->attrdef = NULL;
2231 iput_upcase_err_out:
2232 #endif /* NTFS_RW */
2233 vol->upcase_len = 0;
2234 mutex_lock(&ntfs_lock);
2235 if (vol->upcase == default_upcase) {
2236 ntfs_nr_upcase_users--;
2239 mutex_unlock(&ntfs_lock);
2241 ntfs_free(vol->upcase);
2244 iput_mftbmp_err_out:
2245 iput(vol->mftbmp_ino);
2248 iput(vol->mftmirr_ino);
2249 #endif /* NTFS_RW */
2254 * ntfs_put_super - called by the vfs to unmount a volume
2255 * @sb: vfs superblock of volume to unmount
2257 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2258 * the volume is being unmounted (umount system call has been invoked) and it
2259 * releases all inodes and memory belonging to the NTFS specific part of the
2262 static void ntfs_put_super(struct super_block *sb)
2264 ntfs_volume *vol = NTFS_SB(sb);
2266 ntfs_debug("Entering.");
2270 * Commit all inodes while they are still open in case some of them
2271 * cause others to be dirtied.
2273 ntfs_commit_inode(vol->vol_ino);
2275 /* NTFS 3.0+ specific. */
2276 if (vol->major_ver >= 3) {
2277 if (vol->usnjrnl_j_ino)
2278 ntfs_commit_inode(vol->usnjrnl_j_ino);
2279 if (vol->usnjrnl_max_ino)
2280 ntfs_commit_inode(vol->usnjrnl_max_ino);
2281 if (vol->usnjrnl_ino)
2282 ntfs_commit_inode(vol->usnjrnl_ino);
2283 if (vol->quota_q_ino)
2284 ntfs_commit_inode(vol->quota_q_ino);
2286 ntfs_commit_inode(vol->quota_ino);
2287 if (vol->extend_ino)
2288 ntfs_commit_inode(vol->extend_ino);
2289 if (vol->secure_ino)
2290 ntfs_commit_inode(vol->secure_ino);
2293 ntfs_commit_inode(vol->root_ino);
2295 down_write(&vol->lcnbmp_lock);
2296 ntfs_commit_inode(vol->lcnbmp_ino);
2297 up_write(&vol->lcnbmp_lock);
2299 down_write(&vol->mftbmp_lock);
2300 ntfs_commit_inode(vol->mftbmp_ino);
2301 up_write(&vol->mftbmp_lock);
2303 if (vol->logfile_ino)
2304 ntfs_commit_inode(vol->logfile_ino);
2306 if (vol->mftmirr_ino)
2307 ntfs_commit_inode(vol->mftmirr_ino);
2308 ntfs_commit_inode(vol->mft_ino);
2311 * If a read-write mount and no volume errors have occurred, mark the
2312 * volume clean. Also, re-commit all affected inodes.
2314 if (!sb_rdonly(sb)) {
2315 if (!NVolErrors(vol)) {
2316 if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2317 ntfs_warning(sb, "Failed to clear dirty bit "
2318 "in volume information "
2319 "flags. Run chkdsk.");
2320 ntfs_commit_inode(vol->vol_ino);
2321 ntfs_commit_inode(vol->root_ino);
2322 if (vol->mftmirr_ino)
2323 ntfs_commit_inode(vol->mftmirr_ino);
2324 ntfs_commit_inode(vol->mft_ino);
2326 ntfs_warning(sb, "Volume has errors. Leaving volume "
2327 "marked dirty. Run chkdsk.");
2330 #endif /* NTFS_RW */
2333 vol->vol_ino = NULL;
2335 /* NTFS 3.0+ specific clean up. */
2336 if (vol->major_ver >= 3) {
2338 if (vol->usnjrnl_j_ino) {
2339 iput(vol->usnjrnl_j_ino);
2340 vol->usnjrnl_j_ino = NULL;
2342 if (vol->usnjrnl_max_ino) {
2343 iput(vol->usnjrnl_max_ino);
2344 vol->usnjrnl_max_ino = NULL;
2346 if (vol->usnjrnl_ino) {
2347 iput(vol->usnjrnl_ino);
2348 vol->usnjrnl_ino = NULL;
2350 if (vol->quota_q_ino) {
2351 iput(vol->quota_q_ino);
2352 vol->quota_q_ino = NULL;
2354 if (vol->quota_ino) {
2355 iput(vol->quota_ino);
2356 vol->quota_ino = NULL;
2358 #endif /* NTFS_RW */
2359 if (vol->extend_ino) {
2360 iput(vol->extend_ino);
2361 vol->extend_ino = NULL;
2363 if (vol->secure_ino) {
2364 iput(vol->secure_ino);
2365 vol->secure_ino = NULL;
2369 iput(vol->root_ino);
2370 vol->root_ino = NULL;
2372 down_write(&vol->lcnbmp_lock);
2373 iput(vol->lcnbmp_ino);
2374 vol->lcnbmp_ino = NULL;
2375 up_write(&vol->lcnbmp_lock);
2377 down_write(&vol->mftbmp_lock);
2378 iput(vol->mftbmp_ino);
2379 vol->mftbmp_ino = NULL;
2380 up_write(&vol->mftbmp_lock);
2383 if (vol->logfile_ino) {
2384 iput(vol->logfile_ino);
2385 vol->logfile_ino = NULL;
2387 if (vol->mftmirr_ino) {
2388 /* Re-commit the mft mirror and mft just in case. */
2389 ntfs_commit_inode(vol->mftmirr_ino);
2390 ntfs_commit_inode(vol->mft_ino);
2391 iput(vol->mftmirr_ino);
2392 vol->mftmirr_ino = NULL;
2395 * We should have no dirty inodes left, due to
2396 * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2397 * the underlying mft records are written out and cleaned.
2399 ntfs_commit_inode(vol->mft_ino);
2400 write_inode_now(vol->mft_ino, 1);
2401 #endif /* NTFS_RW */
2404 vol->mft_ino = NULL;
2406 /* Throw away the table of attribute definitions. */
2407 vol->attrdef_size = 0;
2409 ntfs_free(vol->attrdef);
2410 vol->attrdef = NULL;
2412 vol->upcase_len = 0;
2414 * Destroy the global default upcase table if necessary. Also decrease
2415 * the number of upcase users if we are a user.
2417 mutex_lock(&ntfs_lock);
2418 if (vol->upcase == default_upcase) {
2419 ntfs_nr_upcase_users--;
2422 if (!ntfs_nr_upcase_users && default_upcase) {
2423 ntfs_free(default_upcase);
2424 default_upcase = NULL;
2426 if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2427 free_compression_buffers();
2428 mutex_unlock(&ntfs_lock);
2430 ntfs_free(vol->upcase);
2434 unload_nls(vol->nls_map);
2436 sb->s_fs_info = NULL;
2441 * get_nr_free_clusters - return the number of free clusters on a volume
2442 * @vol: ntfs volume for which to obtain free cluster count
2444 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2445 * actually calculate the number of clusters in use instead because this
2446 * allows us to not care about partial pages as these will be just zero filled
2447 * and hence not be counted as allocated clusters.
2449 * The only particularity is that clusters beyond the end of the logical ntfs
2450 * volume will be marked as allocated to prevent errors which means we have to
2451 * discount those at the end. This is important as the cluster bitmap always
2452 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2453 * the logical volume and marked in use when they are not as they do not exist.
2455 * If any pages cannot be read we assume all clusters in the erroring pages are
2456 * in use. This means we return an underestimate on errors which is better than
2459 static s64 get_nr_free_clusters(ntfs_volume *vol)
2461 s64 nr_free = vol->nr_clusters;
2462 struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2464 pgoff_t index, max_index;
2466 ntfs_debug("Entering.");
2467 /* Serialize accesses to the cluster bitmap. */
2468 down_read(&vol->lcnbmp_lock);
2470 * Convert the number of bits into bytes rounded up, then convert into
2471 * multiples of PAGE_SIZE, rounding up so that if we have one
2472 * full and one partial page max_index = 2.
2474 max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_SIZE - 1) >>
2476 /* Use multiples of 4 bytes, thus max_size is PAGE_SIZE / 4. */
2477 ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2478 max_index, PAGE_SIZE / 4);
2479 for (index = 0; index < max_index; index++) {
2480 unsigned long *kaddr;
2483 * Read the page from page cache, getting it from backing store
2484 * if necessary, and increment the use count.
2486 page = read_mapping_page(mapping, index, NULL);
2487 /* Ignore pages which errored synchronously. */
2489 ntfs_debug("read_mapping_page() error. Skipping "
2490 "page (index 0x%lx).", index);
2491 nr_free -= PAGE_SIZE * 8;
2494 kaddr = kmap_atomic(page);
2496 * Subtract the number of set bits. If this
2497 * is the last page and it is partial we don't really care as
2498 * it just means we do a little extra work but it won't affect
2499 * the result as all out of range bytes are set to zero by
2502 nr_free -= bitmap_weight(kaddr,
2503 PAGE_SIZE * BITS_PER_BYTE);
2504 kunmap_atomic(kaddr);
2507 ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2509 * Fixup for eventual bits outside logical ntfs volume (see function
2510 * description above).
2512 if (vol->nr_clusters & 63)
2513 nr_free += 64 - (vol->nr_clusters & 63);
2514 up_read(&vol->lcnbmp_lock);
2515 /* If errors occurred we may well have gone below zero, fix this. */
2518 ntfs_debug("Exiting.");
2523 * __get_nr_free_mft_records - return the number of free inodes on a volume
2524 * @vol: ntfs volume for which to obtain free inode count
2525 * @nr_free: number of mft records in filesystem
2526 * @max_index: maximum number of pages containing set bits
2528 * Calculate the number of free mft records (inodes) on the mounted NTFS
2529 * volume @vol. We actually calculate the number of mft records in use instead
2530 * because this allows us to not care about partial pages as these will be just
2531 * zero filled and hence not be counted as allocated mft record.
2533 * If any pages cannot be read we assume all mft records in the erroring pages
2534 * are in use. This means we return an underestimate on errors which is better
2535 * than an overestimate.
2537 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2539 static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2540 s64 nr_free, const pgoff_t max_index)
2542 struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2546 ntfs_debug("Entering.");
2547 /* Use multiples of 4 bytes, thus max_size is PAGE_SIZE / 4. */
2548 ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2549 "0x%lx.", max_index, PAGE_SIZE / 4);
2550 for (index = 0; index < max_index; index++) {
2551 unsigned long *kaddr;
2554 * Read the page from page cache, getting it from backing store
2555 * if necessary, and increment the use count.
2557 page = read_mapping_page(mapping, index, NULL);
2558 /* Ignore pages which errored synchronously. */
2560 ntfs_debug("read_mapping_page() error. Skipping "
2561 "page (index 0x%lx).", index);
2562 nr_free -= PAGE_SIZE * 8;
2565 kaddr = kmap_atomic(page);
2567 * Subtract the number of set bits. If this
2568 * is the last page and it is partial we don't really care as
2569 * it just means we do a little extra work but it won't affect
2570 * the result as all out of range bytes are set to zero by
2573 nr_free -= bitmap_weight(kaddr,
2574 PAGE_SIZE * BITS_PER_BYTE);
2575 kunmap_atomic(kaddr);
2578 ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2580 /* If errors occurred we may well have gone below zero, fix this. */
2583 ntfs_debug("Exiting.");
2588 * ntfs_statfs - return information about mounted NTFS volume
2589 * @dentry: dentry from mounted volume
2590 * @sfs: statfs structure in which to return the information
2592 * Return information about the mounted NTFS volume @dentry in the statfs structure
2593 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2594 * called). We interpret the values to be correct of the moment in time at
2595 * which we are called. Most values are variable otherwise and this isn't just
2596 * the free values but the totals as well. For example we can increase the
2597 * total number of file nodes if we run out and we can keep doing this until
2598 * there is no more space on the volume left at all.
2600 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2601 * ustat system calls.
2603 * Return 0 on success or -errno on error.
2605 static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2607 struct super_block *sb = dentry->d_sb;
2609 ntfs_volume *vol = NTFS_SB(sb);
2610 ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2612 unsigned long flags;
2614 ntfs_debug("Entering.");
2615 /* Type of filesystem. */
2616 sfs->f_type = NTFS_SB_MAGIC;
2617 /* Optimal transfer block size. */
2618 sfs->f_bsize = PAGE_SIZE;
2620 * Total data blocks in filesystem in units of f_bsize and since
2621 * inodes are also stored in data blocs ($MFT is a file) this is just
2622 * the total clusters.
2624 sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2626 /* Free data blocks in filesystem in units of f_bsize. */
2627 size = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2631 /* Free blocks avail to non-superuser, same as above on NTFS. */
2632 sfs->f_bavail = sfs->f_bfree = size;
2633 /* Serialize accesses to the inode bitmap. */
2634 down_read(&vol->mftbmp_lock);
2635 read_lock_irqsave(&mft_ni->size_lock, flags);
2636 size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2638 * Convert the maximum number of set bits into bytes rounded up, then
2639 * convert into multiples of PAGE_SIZE, rounding up so that if we
2640 * have one full and one partial page max_index = 2.
2642 max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2643 + 7) >> 3) + PAGE_SIZE - 1) >> PAGE_SHIFT;
2644 read_unlock_irqrestore(&mft_ni->size_lock, flags);
2645 /* Number of inodes in filesystem (at this point in time). */
2646 sfs->f_files = size;
2647 /* Free inodes in fs (based on current total count). */
2648 sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2649 up_read(&vol->mftbmp_lock);
2651 * File system id. This is extremely *nix flavour dependent and even
2652 * within Linux itself all fs do their own thing. I interpret this to
2653 * mean a unique id associated with the mounted fs and not the id
2654 * associated with the filesystem driver, the latter is already given
2655 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2656 * volume serial number splitting it into two 32-bit parts. We enter
2657 * the least significant 32-bits in f_fsid[0] and the most significant
2658 * 32-bits in f_fsid[1].
2660 sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2661 sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2662 /* Maximum length of filenames. */
2663 sfs->f_namelen = NTFS_MAX_NAME_LEN;
2668 static int ntfs_write_inode(struct inode *vi, struct writeback_control *wbc)
2670 return __ntfs_write_inode(vi, wbc->sync_mode == WB_SYNC_ALL);
2675 * The complete super operations.
2677 static const struct super_operations ntfs_sops = {
2678 .alloc_inode = ntfs_alloc_big_inode, /* VFS: Allocate new inode. */
2679 .free_inode = ntfs_free_big_inode, /* VFS: Deallocate inode. */
2681 .write_inode = ntfs_write_inode, /* VFS: Write dirty inode to
2683 #endif /* NTFS_RW */
2684 .put_super = ntfs_put_super, /* Syscall: umount. */
2685 .statfs = ntfs_statfs, /* Syscall: statfs */
2686 .remount_fs = ntfs_remount, /* Syscall: mount -o remount. */
2687 .evict_inode = ntfs_evict_big_inode, /* VFS: Called when an inode is
2688 removed from memory. */
2689 .show_options = ntfs_show_options, /* Show mount options in
2694 * ntfs_fill_super - mount an ntfs filesystem
2695 * @sb: super block of ntfs filesystem to mount
2696 * @opt: string containing the mount options
2697 * @silent: silence error output
2699 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2700 * with the mount otions in @data with the NTFS filesystem.
2702 * If @silent is true, remain silent even if errors are detected. This is used
2703 * during bootup, when the kernel tries to mount the root filesystem with all
2704 * registered filesystems one after the other until one succeeds. This implies
2705 * that all filesystems except the correct one will quite correctly and
2706 * expectedly return an error, but nobody wants to see error messages when in
2707 * fact this is what is supposed to happen.
2709 * NOTE: @sb->s_flags contains the mount options flags.
2711 static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2714 struct buffer_head *bh;
2715 struct inode *tmp_ino;
2716 int blocksize, result;
2719 * We do a pretty difficult piece of bootstrap by reading the
2720 * MFT (and other metadata) from disk into memory. We'll only
2721 * release this metadata during umount, so the locking patterns
2722 * observed during bootstrap do not count. So turn off the
2723 * observation of locking patterns (strictly for this context
2724 * only) while mounting NTFS. [The validator is still active
2725 * otherwise, even for this context: it will for example record
2726 * lock class registrations.]
2729 ntfs_debug("Entering.");
2731 sb->s_flags |= SB_RDONLY;
2732 #endif /* ! NTFS_RW */
2733 /* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2734 sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2738 ntfs_error(sb, "Allocation of NTFS volume structure "
2739 "failed. Aborting mount...");
2743 /* Initialize ntfs_volume structure. */
2744 *vol = (ntfs_volume) {
2747 * Default is group and other don't have any access to files or
2748 * directories while owner has full access. Further, files by
2749 * default are not executable but directories are of course
2755 init_rwsem(&vol->mftbmp_lock);
2756 init_rwsem(&vol->lcnbmp_lock);
2758 /* By default, enable sparse support. */
2759 NVolSetSparseEnabled(vol);
2761 /* Important to get the mount options dealt with now. */
2762 if (!parse_options(vol, (char*)opt))
2765 /* We support sector sizes up to the PAGE_SIZE. */
2766 if (bdev_logical_block_size(sb->s_bdev) > PAGE_SIZE) {
2768 ntfs_error(sb, "Device has unsupported sector size "
2769 "(%i). The maximum supported sector "
2770 "size on this architecture is %lu "
2772 bdev_logical_block_size(sb->s_bdev),
2777 * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2778 * sector size, whichever is bigger.
2780 blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2781 if (blocksize < NTFS_BLOCK_SIZE) {
2783 ntfs_error(sb, "Unable to set device block size.");
2786 BUG_ON(blocksize != sb->s_blocksize);
2787 ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2788 blocksize, sb->s_blocksize_bits);
2789 /* Determine the size of the device in units of block_size bytes. */
2790 if (!i_size_read(sb->s_bdev->bd_inode)) {
2792 ntfs_error(sb, "Unable to determine device size.");
2795 vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2796 sb->s_blocksize_bits;
2797 /* Read the boot sector and return unlocked buffer head to it. */
2798 if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2800 ntfs_error(sb, "Not an NTFS volume.");
2804 * Extract the data from the boot sector and setup the ntfs volume
2807 result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2811 ntfs_error(sb, "Unsupported NTFS filesystem.");
2815 * If the boot sector indicates a sector size bigger than the current
2816 * device block size, switch the device block size to the sector size.
2817 * TODO: It may be possible to support this case even when the set
2818 * below fails, we would just be breaking up the i/o for each sector
2819 * into multiple blocks for i/o purposes but otherwise it should just
2820 * work. However it is safer to leave disabled until someone hits this
2821 * error message and then we can get them to try it without the setting
2822 * so we know for sure that it works.
2824 if (vol->sector_size > blocksize) {
2825 blocksize = sb_set_blocksize(sb, vol->sector_size);
2826 if (blocksize != vol->sector_size) {
2828 ntfs_error(sb, "Unable to set device block "
2829 "size to sector size (%i).",
2833 BUG_ON(blocksize != sb->s_blocksize);
2834 vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2835 sb->s_blocksize_bits;
2836 ntfs_debug("Changed device block size to %i bytes (block size "
2837 "bits %i) to match volume sector size.",
2838 blocksize, sb->s_blocksize_bits);
2840 /* Initialize the cluster and mft allocators. */
2841 ntfs_setup_allocators(vol);
2842 /* Setup remaining fields in the super block. */
2843 sb->s_magic = NTFS_SB_MAGIC;
2845 * Ntfs allows 63 bits for the file size, i.e. correct would be:
2846 * sb->s_maxbytes = ~0ULL >> 1;
2847 * But the kernel uses a long as the page cache page index which on
2848 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2849 * defined to the maximum the page cache page index can cope with
2850 * without overflowing the index or to 2^63 - 1, whichever is smaller.
2852 sb->s_maxbytes = MAX_LFS_FILESIZE;
2853 /* Ntfs measures time in 100ns intervals. */
2854 sb->s_time_gran = 100;
2856 * Now load the metadata required for the page cache and our address
2857 * space operations to function. We do this by setting up a specialised
2858 * read_inode method and then just calling the normal iget() to obtain
2859 * the inode for $MFT which is sufficient to allow our normal inode
2860 * operations and associated address space operations to function.
2862 sb->s_op = &ntfs_sops;
2863 tmp_ino = new_inode(sb);
2866 ntfs_error(sb, "Failed to load essential metadata.");
2869 tmp_ino->i_ino = FILE_MFT;
2870 insert_inode_hash(tmp_ino);
2871 if (ntfs_read_inode_mount(tmp_ino) < 0) {
2873 ntfs_error(sb, "Failed to load essential metadata.");
2874 goto iput_tmp_ino_err_out_now;
2876 mutex_lock(&ntfs_lock);
2878 * The current mount is a compression user if the cluster size is
2879 * less than or equal 4kiB.
2881 if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2882 result = allocate_compression_buffers();
2884 ntfs_error(NULL, "Failed to allocate buffers "
2885 "for compression engine.");
2886 ntfs_nr_compression_users--;
2887 mutex_unlock(&ntfs_lock);
2888 goto iput_tmp_ino_err_out_now;
2892 * Generate the global default upcase table if necessary. Also
2893 * temporarily increment the number of upcase users to avoid race
2894 * conditions with concurrent (u)mounts.
2896 if (!default_upcase)
2897 default_upcase = generate_default_upcase();
2898 ntfs_nr_upcase_users++;
2899 mutex_unlock(&ntfs_lock);
2901 * From now on, ignore @silent parameter. If we fail below this line,
2902 * it will be due to a corrupt fs or a system error, so we report it.
2905 * Open the system files with normal access functions and complete
2906 * setting up the ntfs super block.
2908 if (!load_system_files(vol)) {
2909 ntfs_error(sb, "Failed to load system files.");
2910 goto unl_upcase_iput_tmp_ino_err_out_now;
2913 /* We grab a reference, simulating an ntfs_iget(). */
2914 ihold(vol->root_ino);
2915 if ((sb->s_root = d_make_root(vol->root_ino))) {
2916 ntfs_debug("Exiting, status successful.");
2917 /* Release the default upcase if it has no users. */
2918 mutex_lock(&ntfs_lock);
2919 if (!--ntfs_nr_upcase_users && default_upcase) {
2920 ntfs_free(default_upcase);
2921 default_upcase = NULL;
2923 mutex_unlock(&ntfs_lock);
2924 sb->s_export_op = &ntfs_export_ops;
2928 ntfs_error(sb, "Failed to allocate root directory.");
2929 /* Clean up after the successful load_system_files() call from above. */
2930 // TODO: Use ntfs_put_super() instead of repeating all this code...
2931 // FIXME: Should mark the volume clean as the error is most likely
2934 vol->vol_ino = NULL;
2935 /* NTFS 3.0+ specific clean up. */
2936 if (vol->major_ver >= 3) {
2938 if (vol->usnjrnl_j_ino) {
2939 iput(vol->usnjrnl_j_ino);
2940 vol->usnjrnl_j_ino = NULL;
2942 if (vol->usnjrnl_max_ino) {
2943 iput(vol->usnjrnl_max_ino);
2944 vol->usnjrnl_max_ino = NULL;
2946 if (vol->usnjrnl_ino) {
2947 iput(vol->usnjrnl_ino);
2948 vol->usnjrnl_ino = NULL;
2950 if (vol->quota_q_ino) {
2951 iput(vol->quota_q_ino);
2952 vol->quota_q_ino = NULL;
2954 if (vol->quota_ino) {
2955 iput(vol->quota_ino);
2956 vol->quota_ino = NULL;
2958 #endif /* NTFS_RW */
2959 if (vol->extend_ino) {
2960 iput(vol->extend_ino);
2961 vol->extend_ino = NULL;
2963 if (vol->secure_ino) {
2964 iput(vol->secure_ino);
2965 vol->secure_ino = NULL;
2968 iput(vol->root_ino);
2969 vol->root_ino = NULL;
2970 iput(vol->lcnbmp_ino);
2971 vol->lcnbmp_ino = NULL;
2972 iput(vol->mftbmp_ino);
2973 vol->mftbmp_ino = NULL;
2975 if (vol->logfile_ino) {
2976 iput(vol->logfile_ino);
2977 vol->logfile_ino = NULL;
2979 if (vol->mftmirr_ino) {
2980 iput(vol->mftmirr_ino);
2981 vol->mftmirr_ino = NULL;
2983 #endif /* NTFS_RW */
2984 /* Throw away the table of attribute definitions. */
2985 vol->attrdef_size = 0;
2987 ntfs_free(vol->attrdef);
2988 vol->attrdef = NULL;
2990 vol->upcase_len = 0;
2991 mutex_lock(&ntfs_lock);
2992 if (vol->upcase == default_upcase) {
2993 ntfs_nr_upcase_users--;
2996 mutex_unlock(&ntfs_lock);
2998 ntfs_free(vol->upcase);
3002 unload_nls(vol->nls_map);
3003 vol->nls_map = NULL;
3005 /* Error exit code path. */
3006 unl_upcase_iput_tmp_ino_err_out_now:
3008 * Decrease the number of upcase users and destroy the global default
3009 * upcase table if necessary.
3011 mutex_lock(&ntfs_lock);
3012 if (!--ntfs_nr_upcase_users && default_upcase) {
3013 ntfs_free(default_upcase);
3014 default_upcase = NULL;
3016 if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
3017 free_compression_buffers();
3018 mutex_unlock(&ntfs_lock);
3019 iput_tmp_ino_err_out_now:
3021 if (vol->mft_ino && vol->mft_ino != tmp_ino)
3023 vol->mft_ino = NULL;
3024 /* Errors at this stage are irrelevant. */
3026 sb->s_fs_info = NULL;
3028 ntfs_debug("Failed, returning -EINVAL.");
3034 * This is a slab cache to optimize allocations and deallocations of Unicode
3035 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3036 * (255) Unicode characters + a terminating NULL Unicode character.
3038 struct kmem_cache *ntfs_name_cache;
3040 /* Slab caches for efficient allocation/deallocation of inodes. */
3041 struct kmem_cache *ntfs_inode_cache;
3042 struct kmem_cache *ntfs_big_inode_cache;
3044 /* Init once constructor for the inode slab cache. */
3045 static void ntfs_big_inode_init_once(void *foo)
3047 ntfs_inode *ni = (ntfs_inode *)foo;
3049 inode_init_once(VFS_I(ni));
3053 * Slab caches to optimize allocations and deallocations of attribute search
3054 * contexts and index contexts, respectively.
3056 struct kmem_cache *ntfs_attr_ctx_cache;
3057 struct kmem_cache *ntfs_index_ctx_cache;
3059 /* Driver wide mutex. */
3060 DEFINE_MUTEX(ntfs_lock);
3062 static struct dentry *ntfs_mount(struct file_system_type *fs_type,
3063 int flags, const char *dev_name, void *data)
3065 return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
3068 static struct file_system_type ntfs_fs_type = {
3069 .owner = THIS_MODULE,
3071 .mount = ntfs_mount,
3072 .kill_sb = kill_block_super,
3073 .fs_flags = FS_REQUIRES_DEV,
3075 MODULE_ALIAS_FS("ntfs");
3077 /* Stable names for the slab caches. */
3078 static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3079 static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3080 static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3081 static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3082 static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3084 static int __init init_ntfs_fs(void)
3088 /* This may be ugly but it results in pretty output so who cares. (-8 */
3089 pr_info("driver " NTFS_VERSION " [Flags: R/"
3103 ntfs_debug("Debug messages are enabled.");
3105 ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3106 sizeof(ntfs_index_context), 0 /* offset */,
3107 SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3108 if (!ntfs_index_ctx_cache) {
3109 pr_crit("Failed to create %s!\n", ntfs_index_ctx_cache_name);
3112 ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3113 sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3114 SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3115 if (!ntfs_attr_ctx_cache) {
3116 pr_crit("NTFS: Failed to create %s!\n",
3117 ntfs_attr_ctx_cache_name);
3121 ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3122 (NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3123 SLAB_HWCACHE_ALIGN, NULL);
3124 if (!ntfs_name_cache) {
3125 pr_crit("Failed to create %s!\n", ntfs_name_cache_name);
3129 ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3130 sizeof(ntfs_inode), 0,
3131 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
3132 if (!ntfs_inode_cache) {
3133 pr_crit("Failed to create %s!\n", ntfs_inode_cache_name);
3137 ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3138 sizeof(big_ntfs_inode), 0,
3139 SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD|
3140 SLAB_ACCOUNT, ntfs_big_inode_init_once);
3141 if (!ntfs_big_inode_cache) {
3142 pr_crit("Failed to create %s!\n", ntfs_big_inode_cache_name);
3143 goto big_inode_err_out;
3146 /* Register the ntfs sysctls. */
3147 err = ntfs_sysctl(1);
3149 pr_crit("Failed to register NTFS sysctls!\n");
3150 goto sysctl_err_out;
3153 err = register_filesystem(&ntfs_fs_type);
3155 ntfs_debug("NTFS driver registered successfully.");
3156 return 0; /* Success! */
3158 pr_crit("Failed to register NTFS filesystem driver!\n");
3160 /* Unregister the ntfs sysctls. */
3163 kmem_cache_destroy(ntfs_big_inode_cache);
3165 kmem_cache_destroy(ntfs_inode_cache);
3167 kmem_cache_destroy(ntfs_name_cache);
3169 kmem_cache_destroy(ntfs_attr_ctx_cache);
3171 kmem_cache_destroy(ntfs_index_ctx_cache);
3174 pr_crit("Aborting NTFS filesystem driver registration...\n");
3180 static void __exit exit_ntfs_fs(void)
3182 ntfs_debug("Unregistering NTFS driver.");
3184 unregister_filesystem(&ntfs_fs_type);
3187 * Make sure all delayed rcu free inodes are flushed before we
3191 kmem_cache_destroy(ntfs_big_inode_cache);
3192 kmem_cache_destroy(ntfs_inode_cache);
3193 kmem_cache_destroy(ntfs_name_cache);
3194 kmem_cache_destroy(ntfs_attr_ctx_cache);
3195 kmem_cache_destroy(ntfs_index_ctx_cache);
3196 /* Unregister the ntfs sysctls. */
3200 MODULE_AUTHOR("Anton Altaparmakov <anton@tuxera.com>");
3201 MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2014 Anton Altaparmakov and Tuxera Inc.");
3202 MODULE_VERSION(NTFS_VERSION);
3203 MODULE_LICENSE("GPL");
3205 module_param(debug_msgs, bint, 0);
3206 MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3209 module_init(init_ntfs_fs)
3210 module_exit(exit_ntfs_fs)
git.samba.org / sfrench / cifs-2.6.git / blob