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
23 #include <linux/stddef.h>
24 #include <linux/init.h>
25 #include <linux/slab.h>
26 #include <linux/string.h>
27 #include <linux/spinlock.h>
28 #include <linux/blkdev.h> /* For bdev_logical_block_size(). */
29 #include <linux/backing-dev.h>
30 #include <linux/buffer_head.h>
31 #include <linux/vfs.h>
32 #include <linux/moduleparam.h>
33 #include <linux/bitmap.h>
48 /* Number of mounted filesystems which have compression enabled. */
49 static unsigned long ntfs_nr_compression_users;
51 /* A global default upcase table and a corresponding reference count. */
52 static ntfschar *default_upcase = NULL;
53 static unsigned long ntfs_nr_upcase_users = 0;
55 /* Error constants/strings used in inode.c::ntfs_show_options(). */
57 /* One of these must be present, default is ON_ERRORS_CONTINUE. */
58 ON_ERRORS_PANIC = 0x01,
59 ON_ERRORS_REMOUNT_RO = 0x02,
60 ON_ERRORS_CONTINUE = 0x04,
61 /* Optional, can be combined with any of the above. */
62 ON_ERRORS_RECOVER = 0x10,
65 const option_t on_errors_arr[] = {
66 { ON_ERRORS_PANIC, "panic" },
67 { ON_ERRORS_REMOUNT_RO, "remount-ro", },
68 { ON_ERRORS_CONTINUE, "continue", },
69 { ON_ERRORS_RECOVER, "recover" },
76 * Copied from old ntfs driver (which copied from vfat driver).
78 static int simple_getbool(char *s, bool *setval)
81 if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
83 else if (!strcmp(s, "0") || !strcmp(s, "no") ||
94 * parse_options - parse the (re)mount options
96 * @opt: string containing the (re)mount options
98 * Parse the recognized options in @opt for the ntfs volume described by @vol.
100 static bool parse_options(ntfs_volume *vol, char *opt)
103 static char *utf8 = "utf8";
104 int errors = 0, sloppy = 0;
105 kuid_t uid = INVALID_UID;
106 kgid_t gid = INVALID_GID;
107 umode_t fmask = (umode_t)-1, dmask = (umode_t)-1;
108 int mft_zone_multiplier = -1, on_errors = -1;
109 int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
110 struct nls_table *nls_map = NULL, *old_nls;
112 /* I am lazy... (-8 */
113 #define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value) \
114 if (!strcmp(p, option)) { \
116 variable = default_value; \
118 variable = simple_strtoul(ov = v, &v, 0); \
123 #define NTFS_GETOPT(option, variable) \
124 if (!strcmp(p, option)) { \
127 variable = simple_strtoul(ov = v, &v, 0); \
131 #define NTFS_GETOPT_UID(option, variable) \
132 if (!strcmp(p, option)) { \
136 uid_value = simple_strtoul(ov = v, &v, 0); \
139 variable = make_kuid(current_user_ns(), uid_value); \
140 if (!uid_valid(variable)) \
143 #define NTFS_GETOPT_GID(option, variable) \
144 if (!strcmp(p, option)) { \
148 gid_value = simple_strtoul(ov = v, &v, 0); \
151 variable = make_kgid(current_user_ns(), gid_value); \
152 if (!gid_valid(variable)) \
155 #define NTFS_GETOPT_OCTAL(option, variable) \
156 if (!strcmp(p, option)) { \
159 variable = simple_strtoul(ov = v, &v, 8); \
163 #define NTFS_GETOPT_BOOL(option, variable) \
164 if (!strcmp(p, option)) { \
166 if (!simple_getbool(v, &val)) \
170 #define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array) \
171 if (!strcmp(p, option)) { \
176 if (variable == -1) \
178 for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
179 if (!strcmp(opt_array[_i].str, v)) { \
180 variable |= opt_array[_i].val; \
183 if (!opt_array[_i].str || !*opt_array[_i].str) \
187 goto no_mount_options;
188 ntfs_debug("Entering with mount options string: %s", opt);
189 while ((p = strsep(&opt, ","))) {
190 if ((v = strchr(p, '=')))
192 NTFS_GETOPT_UID("uid", uid)
193 else NTFS_GETOPT_GID("gid", gid)
194 else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
195 else NTFS_GETOPT_OCTAL("fmask", fmask)
196 else NTFS_GETOPT_OCTAL("dmask", dmask)
197 else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
198 else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
199 else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
200 else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
201 else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
202 else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
204 else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
205 ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
207 else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
208 if (!strcmp(p, "iocharset"))
209 ntfs_warning(vol->sb, "Option iocharset is "
210 "deprecated. Please use "
211 "option nls=<charsetname> in "
217 nls_map = load_nls(v);
220 ntfs_error(vol->sb, "NLS character set "
224 ntfs_error(vol->sb, "NLS character set %s not "
225 "found. Using previous one %s.",
226 v, old_nls->charset);
228 } else /* nls_map */ {
231 } else if (!strcmp(p, "utf8")) {
233 ntfs_warning(vol->sb, "Option utf8 is no longer "
234 "supported, using option nls=utf8. Please "
235 "use option nls=utf8 in the future and "
236 "make sure utf8 is compiled either as a "
237 "module or into the kernel.");
240 else if (!simple_getbool(v, &val))
247 ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
248 if (errors < INT_MAX)
251 #undef NTFS_GETOPT_OPTIONS_ARRAY
252 #undef NTFS_GETOPT_BOOL
254 #undef NTFS_GETOPT_WITH_DEFAULT
257 if (errors && !sloppy)
260 ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
261 "unrecognized mount option(s) and continuing.");
262 /* Keep this first! */
263 if (on_errors != -1) {
265 ntfs_error(vol->sb, "Invalid errors option argument "
266 "or bug in options parser.");
271 if (vol->nls_map && vol->nls_map != nls_map) {
272 ntfs_error(vol->sb, "Cannot change NLS character set "
275 } /* else (!vol->nls_map) */
276 ntfs_debug("Using NLS character set %s.", nls_map->charset);
277 vol->nls_map = nls_map;
278 } else /* (!nls_map) */ {
280 vol->nls_map = load_nls_default();
282 ntfs_error(vol->sb, "Failed to load default "
283 "NLS character set.");
286 ntfs_debug("Using default NLS character set (%s).",
287 vol->nls_map->charset);
290 if (mft_zone_multiplier != -1) {
291 if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
292 mft_zone_multiplier) {
293 ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
297 if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
298 ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
299 "Using default value, i.e. 1.");
300 mft_zone_multiplier = 1;
302 vol->mft_zone_multiplier = mft_zone_multiplier;
304 if (!vol->mft_zone_multiplier)
305 vol->mft_zone_multiplier = 1;
307 vol->on_errors = on_errors;
308 if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
309 vol->on_errors |= ON_ERRORS_CONTINUE;
314 if (fmask != (umode_t)-1)
316 if (dmask != (umode_t)-1)
318 if (show_sys_files != -1) {
320 NVolSetShowSystemFiles(vol);
322 NVolClearShowSystemFiles(vol);
324 if (case_sensitive != -1) {
326 NVolSetCaseSensitive(vol);
328 NVolClearCaseSensitive(vol);
330 if (disable_sparse != -1) {
332 NVolClearSparseEnabled(vol);
334 if (!NVolSparseEnabled(vol) &&
335 vol->major_ver && vol->major_ver < 3)
336 ntfs_warning(vol->sb, "Not enabling sparse "
337 "support due to NTFS volume "
338 "version %i.%i (need at least "
339 "version 3.0).", vol->major_ver,
342 NVolSetSparseEnabled(vol);
347 ntfs_error(vol->sb, "The %s option requires an argument.", p);
350 ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
353 ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
360 * ntfs_write_volume_flags - write new flags to the volume information flags
361 * @vol: ntfs volume on which to modify the flags
362 * @flags: new flags value for the volume information flags
364 * Internal function. You probably want to use ntfs_{set,clear}_volume_flags()
365 * instead (see below).
367 * Replace the volume information flags on the volume @vol with the value
368 * supplied in @flags. Note, this overwrites the volume information flags, so
369 * make sure to combine the flags you want to modify with the old flags and use
370 * the result when calling ntfs_write_volume_flags().
372 * Return 0 on success and -errno on error.
374 static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
376 ntfs_inode *ni = NTFS_I(vol->vol_ino);
378 VOLUME_INFORMATION *vi;
379 ntfs_attr_search_ctx *ctx;
382 ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
383 le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
384 if (vol->vol_flags == flags)
387 m = map_mft_record(ni);
392 ctx = ntfs_attr_get_search_ctx(ni, m);
395 goto put_unm_err_out;
397 err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
400 goto put_unm_err_out;
401 vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
402 le16_to_cpu(ctx->attr->data.resident.value_offset));
403 vol->vol_flags = vi->flags = flags;
404 flush_dcache_mft_record_page(ctx->ntfs_ino);
405 mark_mft_record_dirty(ctx->ntfs_ino);
406 ntfs_attr_put_search_ctx(ctx);
407 unmap_mft_record(ni);
413 ntfs_attr_put_search_ctx(ctx);
414 unmap_mft_record(ni);
416 ntfs_error(vol->sb, "Failed with error code %i.", -err);
421 * ntfs_set_volume_flags - set bits in the volume information flags
422 * @vol: ntfs volume on which to modify the flags
423 * @flags: flags to set on the volume
425 * Set the bits in @flags in the volume information flags on the volume @vol.
427 * Return 0 on success and -errno on error.
429 static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
431 flags &= VOLUME_FLAGS_MASK;
432 return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
436 * ntfs_clear_volume_flags - clear bits in the volume information flags
437 * @vol: ntfs volume on which to modify the flags
438 * @flags: flags to clear on the volume
440 * Clear the bits in @flags in the volume information flags on the volume @vol.
442 * Return 0 on success and -errno on error.
444 static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
446 flags &= VOLUME_FLAGS_MASK;
447 flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
448 return ntfs_write_volume_flags(vol, flags);
454 * ntfs_remount - change the mount options of a mounted ntfs filesystem
455 * @sb: superblock of mounted ntfs filesystem
456 * @flags: remount flags
457 * @opt: remount options string
459 * Change the mount options of an already mounted ntfs filesystem.
461 * NOTE: The VFS sets the @sb->s_flags remount flags to @flags after
462 * ntfs_remount() returns successfully (i.e. returns 0). Otherwise,
463 * @sb->s_flags are not changed.
465 static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
467 ntfs_volume *vol = NTFS_SB(sb);
469 ntfs_debug("Entering with remount options string: %s", opt);
472 /* For read-only compiled driver, enforce read-only flag. */
476 * For the read-write compiled driver, if we are remounting read-write,
477 * make sure there are no volume errors and that no unsupported volume
478 * flags are set. Also, empty the logfile journal as it would become
479 * stale as soon as something is written to the volume and mark the
480 * volume dirty so that chkdsk is run if the volume is not umounted
481 * cleanly. Finally, mark the quotas out of date so Windows rescans
482 * the volume on boot and updates them.
484 * When remounting read-only, mark the volume clean if no volume errors
487 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
488 static const char *es = ". Cannot remount read-write.";
490 /* Remounting read-write. */
491 if (NVolErrors(vol)) {
492 ntfs_error(sb, "Volume has errors and is read-only%s",
496 if (vol->vol_flags & VOLUME_IS_DIRTY) {
497 ntfs_error(sb, "Volume is dirty and read-only%s", es);
500 if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
501 ntfs_error(sb, "Volume has been modified by chkdsk "
502 "and is read-only%s", es);
505 if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
506 ntfs_error(sb, "Volume has unsupported flags set "
507 "(0x%x) and is read-only%s",
508 (unsigned)le16_to_cpu(vol->vol_flags),
512 if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
513 ntfs_error(sb, "Failed to set dirty bit in volume "
514 "information flags%s", es);
518 // TODO: Enable this code once we start modifying anything that
519 // is different between NTFS 1.2 and 3.x...
520 /* Set NT4 compatibility flag on newer NTFS version volumes. */
521 if ((vol->major_ver > 1)) {
522 if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
523 ntfs_error(sb, "Failed to set NT4 "
524 "compatibility flag%s", es);
530 if (!ntfs_empty_logfile(vol->logfile_ino)) {
531 ntfs_error(sb, "Failed to empty journal $LogFile%s",
536 if (!ntfs_mark_quotas_out_of_date(vol)) {
537 ntfs_error(sb, "Failed to mark quotas out of date%s",
542 if (!ntfs_stamp_usnjrnl(vol)) {
543 ntfs_error(sb, "Failed to stamp transation log "
548 } else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
549 /* Remounting read-only. */
550 if (!NVolErrors(vol)) {
551 if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
552 ntfs_warning(sb, "Failed to clear dirty bit "
553 "in volume information "
554 "flags. Run chkdsk.");
559 // TODO: Deal with *flags.
561 if (!parse_options(vol, opt))
569 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
570 * @sb: Super block of the device to which @b belongs.
571 * @b: Boot sector of device @sb to check.
572 * @silent: If 'true', all output will be silenced.
574 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
575 * sector. Returns 'true' if it is valid and 'false' if not.
577 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
580 static bool is_boot_sector_ntfs(const struct super_block *sb,
581 const NTFS_BOOT_SECTOR *b, const bool silent)
584 * Check that checksum == sum of u32 values from b to the checksum
585 * field. If checksum is zero, no checking is done. We will work when
586 * the checksum test fails, since some utilities update the boot sector
587 * ignoring the checksum which leaves the checksum out-of-date. We
588 * report a warning if this is the case.
590 if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
594 for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
595 i += le32_to_cpup(u);
596 if (le32_to_cpu(b->checksum) != i)
597 ntfs_warning(sb, "Invalid boot sector checksum.");
599 /* Check OEMidentifier is "NTFS " */
600 if (b->oem_id != magicNTFS)
602 /* Check bytes per sector value is between 256 and 4096. */
603 if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
604 le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
606 /* Check sectors per cluster value is valid. */
607 switch (b->bpb.sectors_per_cluster) {
608 case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
613 /* Check the cluster size is not above the maximum (64kiB). */
614 if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
615 b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
617 /* Check reserved/unused fields are really zero. */
618 if (le16_to_cpu(b->bpb.reserved_sectors) ||
619 le16_to_cpu(b->bpb.root_entries) ||
620 le16_to_cpu(b->bpb.sectors) ||
621 le16_to_cpu(b->bpb.sectors_per_fat) ||
622 le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
624 /* Check clusters per file mft record value is valid. */
625 if ((u8)b->clusters_per_mft_record < 0xe1 ||
626 (u8)b->clusters_per_mft_record > 0xf7)
627 switch (b->clusters_per_mft_record) {
628 case 1: case 2: case 4: case 8: case 16: case 32: case 64:
633 /* Check clusters per index block value is valid. */
634 if ((u8)b->clusters_per_index_record < 0xe1 ||
635 (u8)b->clusters_per_index_record > 0xf7)
636 switch (b->clusters_per_index_record) {
637 case 1: case 2: case 4: case 8: case 16: case 32: case 64:
643 * Check for valid end of sector marker. We will work without it, but
644 * many BIOSes will refuse to boot from a bootsector if the magic is
645 * incorrect, so we emit a warning.
647 if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
648 ntfs_warning(sb, "Invalid end of sector marker.");
655 * read_ntfs_boot_sector - read the NTFS boot sector of a device
656 * @sb: super block of device to read the boot sector from
657 * @silent: if true, suppress all output
659 * Reads the boot sector from the device and validates it. If that fails, tries
660 * to read the backup boot sector, first from the end of the device a-la NT4 and
661 * later and then from the middle of the device a-la NT3.51 and before.
663 * If a valid boot sector is found but it is not the primary boot sector, we
664 * repair the primary boot sector silently (unless the device is read-only or
665 * the primary boot sector is not accessible).
667 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
668 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
669 * to their respective values.
671 * Return the unlocked buffer head containing the boot sector or NULL on error.
673 static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
676 const char *read_err_str = "Unable to read %s boot sector.";
677 struct buffer_head *bh_primary, *bh_backup;
678 sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
680 /* Try to read primary boot sector. */
681 if ((bh_primary = sb_bread(sb, 0))) {
682 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
683 bh_primary->b_data, silent))
686 ntfs_error(sb, "Primary boot sector is invalid.");
688 ntfs_error(sb, read_err_str, "primary");
689 if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
693 ntfs_error(sb, "Mount option errors=recover not used. "
694 "Aborting without trying to recover.");
697 /* Try to read NT4+ backup boot sector. */
698 if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
699 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
700 bh_backup->b_data, silent))
701 goto hotfix_primary_boot_sector;
704 ntfs_error(sb, read_err_str, "backup");
705 /* Try to read NT3.51- backup boot sector. */
706 if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
707 if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
708 bh_backup->b_data, silent))
709 goto hotfix_primary_boot_sector;
711 ntfs_error(sb, "Could not find a valid backup boot "
715 ntfs_error(sb, read_err_str, "backup");
716 /* We failed. Cleanup and return. */
720 hotfix_primary_boot_sector:
723 * If we managed to read sector zero and the volume is not
724 * read-only, copy the found, valid backup boot sector to the
725 * primary boot sector. Note we only copy the actual boot
726 * sector structure, not the actual whole device sector as that
727 * may be bigger and would potentially damage the $Boot system
728 * file (FIXME: Would be nice to know if the backup boot sector
729 * on a large sector device contains the whole boot loader or
730 * just the first 512 bytes).
732 if (!(sb->s_flags & MS_RDONLY)) {
733 ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
734 "boot sector from backup copy.");
735 memcpy(bh_primary->b_data, bh_backup->b_data,
737 mark_buffer_dirty(bh_primary);
738 sync_dirty_buffer(bh_primary);
739 if (buffer_uptodate(bh_primary)) {
743 ntfs_error(sb, "Hot-fix: Device write error while "
744 "recovering primary boot sector.");
746 ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
747 "sector failed: Read-only mount.");
751 ntfs_warning(sb, "Using backup boot sector.");
756 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
757 * @vol: volume structure to initialise with data from boot sector
758 * @b: boot sector to parse
760 * Parse the ntfs boot sector @b and store all imporant information therein in
761 * the ntfs super block @vol. Return 'true' on success and 'false' on error.
763 static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
765 unsigned int sectors_per_cluster_bits, nr_hidden_sects;
766 int clusters_per_mft_record, clusters_per_index_record;
769 vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
770 vol->sector_size_bits = ffs(vol->sector_size) - 1;
771 ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
773 ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
774 vol->sector_size_bits);
775 if (vol->sector_size < vol->sb->s_blocksize) {
776 ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
777 "device block size (%lu). This is not "
778 "supported. Sorry.", vol->sector_size,
779 vol->sb->s_blocksize);
782 ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
783 sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
784 ntfs_debug("sectors_per_cluster_bits = 0x%x",
785 sectors_per_cluster_bits);
786 nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
787 ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
788 vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
789 vol->cluster_size_mask = vol->cluster_size - 1;
790 vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
791 ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
793 ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
794 ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
795 if (vol->cluster_size < vol->sector_size) {
796 ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
797 "sector size (%i). This is not supported. "
798 "Sorry.", vol->cluster_size, vol->sector_size);
801 clusters_per_mft_record = b->clusters_per_mft_record;
802 ntfs_debug("clusters_per_mft_record = %i (0x%x)",
803 clusters_per_mft_record, clusters_per_mft_record);
804 if (clusters_per_mft_record > 0)
805 vol->mft_record_size = vol->cluster_size <<
806 (ffs(clusters_per_mft_record) - 1);
809 * When mft_record_size < cluster_size, clusters_per_mft_record
810 * = -log2(mft_record_size) bytes. mft_record_size normaly is
811 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
813 vol->mft_record_size = 1 << -clusters_per_mft_record;
814 vol->mft_record_size_mask = vol->mft_record_size - 1;
815 vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
816 ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
817 vol->mft_record_size);
818 ntfs_debug("vol->mft_record_size_mask = 0x%x",
819 vol->mft_record_size_mask);
820 ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
821 vol->mft_record_size_bits, vol->mft_record_size_bits);
823 * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
824 * we store $MFT/$DATA, the table of mft records in the page cache.
826 if (vol->mft_record_size > PAGE_CACHE_SIZE) {
827 ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
828 "PAGE_CACHE_SIZE on your system (%lu). "
829 "This is not supported. Sorry.",
830 vol->mft_record_size, PAGE_CACHE_SIZE);
833 /* We cannot support mft record sizes below the sector size. */
834 if (vol->mft_record_size < vol->sector_size) {
835 ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
836 "sector size (%i). This is not supported. "
837 "Sorry.", vol->mft_record_size,
841 clusters_per_index_record = b->clusters_per_index_record;
842 ntfs_debug("clusters_per_index_record = %i (0x%x)",
843 clusters_per_index_record, clusters_per_index_record);
844 if (clusters_per_index_record > 0)
845 vol->index_record_size = vol->cluster_size <<
846 (ffs(clusters_per_index_record) - 1);
849 * When index_record_size < cluster_size,
850 * clusters_per_index_record = -log2(index_record_size) bytes.
851 * index_record_size normaly equals 4096 bytes, which is
852 * encoded as 0xF4 (-12 in decimal).
854 vol->index_record_size = 1 << -clusters_per_index_record;
855 vol->index_record_size_mask = vol->index_record_size - 1;
856 vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
857 ntfs_debug("vol->index_record_size = %i (0x%x)",
858 vol->index_record_size, vol->index_record_size);
859 ntfs_debug("vol->index_record_size_mask = 0x%x",
860 vol->index_record_size_mask);
861 ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
862 vol->index_record_size_bits,
863 vol->index_record_size_bits);
864 /* We cannot support index record sizes below the sector size. */
865 if (vol->index_record_size < vol->sector_size) {
866 ntfs_error(vol->sb, "Index record size (%i) is smaller than "
867 "the sector size (%i). This is not "
868 "supported. Sorry.", vol->index_record_size,
873 * Get the size of the volume in clusters and check for 64-bit-ness.
874 * Windows currently only uses 32 bits to save the clusters so we do
875 * the same as it is much faster on 32-bit CPUs.
877 ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
878 if ((u64)ll >= 1ULL << 32) {
879 ntfs_error(vol->sb, "Cannot handle 64-bit clusters. Sorry.");
882 vol->nr_clusters = ll;
883 ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
885 * On an architecture where unsigned long is 32-bits, we restrict the
886 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
887 * will hopefully optimize the whole check away.
889 if (sizeof(unsigned long) < 8) {
890 if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
891 ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
892 "large for this architecture. "
893 "Maximum supported is 2TiB. Sorry.",
894 (unsigned long long)ll >> (40 -
895 vol->cluster_size_bits));
899 ll = sle64_to_cpu(b->mft_lcn);
900 if (ll >= vol->nr_clusters) {
901 ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
902 "volume. Weird.", (unsigned long long)ll,
903 (unsigned long long)ll);
907 ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
908 ll = sle64_to_cpu(b->mftmirr_lcn);
909 if (ll >= vol->nr_clusters) {
910 ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
911 "of volume. Weird.", (unsigned long long)ll,
912 (unsigned long long)ll);
915 vol->mftmirr_lcn = ll;
916 ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
919 * Work out the size of the mft mirror in number of mft records. If the
920 * cluster size is less than or equal to the size taken by four mft
921 * records, the mft mirror stores the first four mft records. If the
922 * cluster size is bigger than the size taken by four mft records, the
923 * mft mirror contains as many mft records as will fit into one
926 if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
927 vol->mftmirr_size = 4;
929 vol->mftmirr_size = vol->cluster_size >>
930 vol->mft_record_size_bits;
931 ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
933 vol->serial_no = le64_to_cpu(b->volume_serial_number);
934 ntfs_debug("vol->serial_no = 0x%llx",
935 (unsigned long long)vol->serial_no);
940 * ntfs_setup_allocators - initialize the cluster and mft allocators
941 * @vol: volume structure for which to setup the allocators
943 * Setup the cluster (lcn) and mft allocators to the starting values.
945 static void ntfs_setup_allocators(ntfs_volume *vol)
948 LCN mft_zone_size, mft_lcn;
951 ntfs_debug("vol->mft_zone_multiplier = 0x%x",
952 vol->mft_zone_multiplier);
954 /* Determine the size of the MFT zone. */
955 mft_zone_size = vol->nr_clusters;
956 switch (vol->mft_zone_multiplier) { /* % of volume size in clusters */
958 mft_zone_size >>= 1; /* 50% */
961 mft_zone_size = (mft_zone_size +
962 (mft_zone_size >> 1)) >> 2; /* 37.5% */
965 mft_zone_size >>= 2; /* 25% */
969 mft_zone_size >>= 3; /* 12.5% */
972 /* Setup the mft zone. */
973 vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
974 ntfs_debug("vol->mft_zone_pos = 0x%llx",
975 (unsigned long long)vol->mft_zone_pos);
977 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
978 * source) and if the actual mft_lcn is in the expected place or even
979 * further to the front of the volume, extend the mft_zone to cover the
980 * beginning of the volume as well. This is in order to protect the
981 * area reserved for the mft bitmap as well within the mft_zone itself.
982 * On non-standard volumes we do not protect it as the overhead would
983 * be higher than the speed increase we would get by doing it.
985 mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
986 if (mft_lcn * vol->cluster_size < 16 * 1024)
987 mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
989 if (vol->mft_zone_start <= mft_lcn)
990 vol->mft_zone_start = 0;
991 ntfs_debug("vol->mft_zone_start = 0x%llx",
992 (unsigned long long)vol->mft_zone_start);
994 * Need to cap the mft zone on non-standard volumes so that it does
995 * not point outside the boundaries of the volume. We do this by
996 * halving the zone size until we are inside the volume.
998 vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
999 while (vol->mft_zone_end >= vol->nr_clusters) {
1000 mft_zone_size >>= 1;
1001 vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
1003 ntfs_debug("vol->mft_zone_end = 0x%llx",
1004 (unsigned long long)vol->mft_zone_end);
1006 * Set the current position within each data zone to the start of the
1009 vol->data1_zone_pos = vol->mft_zone_end;
1010 ntfs_debug("vol->data1_zone_pos = 0x%llx",
1011 (unsigned long long)vol->data1_zone_pos);
1012 vol->data2_zone_pos = 0;
1013 ntfs_debug("vol->data2_zone_pos = 0x%llx",
1014 (unsigned long long)vol->data2_zone_pos);
1016 /* Set the mft data allocation position to mft record 24. */
1017 vol->mft_data_pos = 24;
1018 ntfs_debug("vol->mft_data_pos = 0x%llx",
1019 (unsigned long long)vol->mft_data_pos);
1020 #endif /* NTFS_RW */
1026 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
1027 * @vol: ntfs super block describing device whose mft mirror to load
1029 * Return 'true' on success or 'false' on error.
1031 static bool load_and_init_mft_mirror(ntfs_volume *vol)
1033 struct inode *tmp_ino;
1036 ntfs_debug("Entering.");
1037 /* Get mft mirror inode. */
1038 tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
1039 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1040 if (!IS_ERR(tmp_ino))
1042 /* Caller will display error message. */
1046 * Re-initialize some specifics about $MFTMirr's inode as
1047 * ntfs_read_inode() will have set up the default ones.
1049 /* Set uid and gid to root. */
1050 tmp_ino->i_uid = GLOBAL_ROOT_UID;
1051 tmp_ino->i_gid = GLOBAL_ROOT_GID;
1052 /* Regular file. No access for anyone. */
1053 tmp_ino->i_mode = S_IFREG;
1054 /* No VFS initiated operations allowed for $MFTMirr. */
1055 tmp_ino->i_op = &ntfs_empty_inode_ops;
1056 tmp_ino->i_fop = &ntfs_empty_file_ops;
1057 /* Put in our special address space operations. */
1058 tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1059 tmp_ni = NTFS_I(tmp_ino);
1060 /* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1061 NInoSetMstProtected(tmp_ni);
1062 NInoSetSparseDisabled(tmp_ni);
1064 * Set up our little cheat allowing us to reuse the async read io
1065 * completion handler for directories.
1067 tmp_ni->itype.index.block_size = vol->mft_record_size;
1068 tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1069 vol->mftmirr_ino = tmp_ino;
1070 ntfs_debug("Done.");
1075 * check_mft_mirror - compare contents of the mft mirror with the mft
1076 * @vol: ntfs super block describing device whose mft mirror to check
1078 * Return 'true' on success or 'false' on error.
1080 * Note, this function also results in the mft mirror runlist being completely
1081 * mapped into memory. The mft mirror write code requires this and will BUG()
1082 * should it find an unmapped runlist element.
1084 static bool check_mft_mirror(ntfs_volume *vol)
1086 struct super_block *sb = vol->sb;
1087 ntfs_inode *mirr_ni;
1088 struct page *mft_page, *mirr_page;
1090 runlist_element *rl, rl2[2];
1092 int mrecs_per_page, i;
1094 ntfs_debug("Entering.");
1095 /* Compare contents of $MFT and $MFTMirr. */
1096 mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1097 BUG_ON(!mrecs_per_page);
1098 BUG_ON(!vol->mftmirr_size);
1099 mft_page = mirr_page = NULL;
1100 kmft = kmirr = NULL;
1105 /* Switch pages if necessary. */
1106 if (!(i % mrecs_per_page)) {
1108 ntfs_unmap_page(mft_page);
1109 ntfs_unmap_page(mirr_page);
1111 /* Get the $MFT page. */
1112 mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1114 if (IS_ERR(mft_page)) {
1115 ntfs_error(sb, "Failed to read $MFT.");
1118 kmft = page_address(mft_page);
1119 /* Get the $MFTMirr page. */
1120 mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1122 if (IS_ERR(mirr_page)) {
1123 ntfs_error(sb, "Failed to read $MFTMirr.");
1126 kmirr = page_address(mirr_page);
1129 /* Do not check the record if it is not in use. */
1130 if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1131 /* Make sure the record is ok. */
1132 if (ntfs_is_baad_recordp((le32*)kmft)) {
1133 ntfs_error(sb, "Incomplete multi sector "
1134 "transfer detected in mft "
1137 ntfs_unmap_page(mirr_page);
1139 ntfs_unmap_page(mft_page);
1143 /* Do not check the mirror record if it is not in use. */
1144 if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1145 if (ntfs_is_baad_recordp((le32*)kmirr)) {
1146 ntfs_error(sb, "Incomplete multi sector "
1147 "transfer detected in mft "
1148 "mirror record %i.", i);
1152 /* Get the amount of data in the current record. */
1153 bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1154 if (bytes < sizeof(MFT_RECORD_OLD) ||
1155 bytes > vol->mft_record_size ||
1156 ntfs_is_baad_recordp((le32*)kmft)) {
1157 bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1158 if (bytes < sizeof(MFT_RECORD_OLD) ||
1159 bytes > vol->mft_record_size ||
1160 ntfs_is_baad_recordp((le32*)kmirr))
1161 bytes = vol->mft_record_size;
1163 /* Compare the two records. */
1164 if (memcmp(kmft, kmirr, bytes)) {
1165 ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1166 "match. Run ntfsfix or chkdsk.", i);
1169 kmft += vol->mft_record_size;
1170 kmirr += vol->mft_record_size;
1171 } while (++i < vol->mftmirr_size);
1172 /* Release the last pages. */
1173 ntfs_unmap_page(mft_page);
1174 ntfs_unmap_page(mirr_page);
1176 /* Construct the mft mirror runlist by hand. */
1178 rl2[0].lcn = vol->mftmirr_lcn;
1179 rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1180 vol->cluster_size - 1) / vol->cluster_size;
1181 rl2[1].vcn = rl2[0].length;
1182 rl2[1].lcn = LCN_ENOENT;
1185 * Because we have just read all of the mft mirror, we know we have
1186 * mapped the full runlist for it.
1188 mirr_ni = NTFS_I(vol->mftmirr_ino);
1189 down_read(&mirr_ni->runlist.lock);
1190 rl = mirr_ni->runlist.rl;
1191 /* Compare the two runlists. They must be identical. */
1194 if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1195 rl2[i].length != rl[i].length) {
1196 ntfs_error(sb, "$MFTMirr location mismatch. "
1198 up_read(&mirr_ni->runlist.lock);
1201 } while (rl2[i++].length);
1202 up_read(&mirr_ni->runlist.lock);
1203 ntfs_debug("Done.");
1208 * load_and_check_logfile - load and check the logfile inode for a volume
1209 * @vol: ntfs super block describing device whose logfile to load
1211 * Return 'true' on success or 'false' on error.
1213 static bool load_and_check_logfile(ntfs_volume *vol,
1214 RESTART_PAGE_HEADER **rp)
1216 struct inode *tmp_ino;
1218 ntfs_debug("Entering.");
1219 tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1220 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1221 if (!IS_ERR(tmp_ino))
1223 /* Caller will display error message. */
1226 if (!ntfs_check_logfile(tmp_ino, rp)) {
1228 /* ntfs_check_logfile() will have displayed error output. */
1231 NInoSetSparseDisabled(NTFS_I(tmp_ino));
1232 vol->logfile_ino = tmp_ino;
1233 ntfs_debug("Done.");
1237 #define NTFS_HIBERFIL_HEADER_SIZE 4096
1240 * check_windows_hibernation_status - check if Windows is suspended on a volume
1241 * @vol: ntfs super block of device to check
1243 * Check if Windows is hibernated on the ntfs volume @vol. This is done by
1244 * looking for the file hiberfil.sys in the root directory of the volume. If
1245 * the file is not present Windows is definitely not suspended.
1247 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1248 * definitely suspended (this volume is not the system volume). Caveat: on a
1249 * system with many volumes it is possible that the < 4kiB check is bogus but
1250 * for now this should do fine.
1252 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1253 * hiberfil header (which is the first 4kiB). If this begins with "hibr",
1254 * Windows is definitely suspended. If it is completely full of zeroes,
1255 * Windows is definitely not hibernated. Any other case is treated as if
1256 * Windows is suspended. This caters for the above mentioned caveat of a
1257 * system with many volumes where no "hibr" magic would be present and there is
1260 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1261 * hibernated on the volume, and -errno on error.
1263 static int check_windows_hibernation_status(ntfs_volume *vol)
1269 ntfs_name *name = NULL;
1271 static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
1272 cpu_to_le16('i'), cpu_to_le16('b'),
1273 cpu_to_le16('e'), cpu_to_le16('r'),
1274 cpu_to_le16('f'), cpu_to_le16('i'),
1275 cpu_to_le16('l'), cpu_to_le16('.'),
1276 cpu_to_le16('s'), cpu_to_le16('y'),
1277 cpu_to_le16('s'), 0 };
1279 ntfs_debug("Entering.");
1281 * Find the inode number for the hibernation file by looking up the
1282 * filename hiberfil.sys in the root directory.
1284 mutex_lock(&vol->root_ino->i_mutex);
1285 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1287 mutex_unlock(&vol->root_ino->i_mutex);
1288 if (IS_ERR_MREF(mref)) {
1289 ret = MREF_ERR(mref);
1290 /* If the file does not exist, Windows is not hibernated. */
1291 if (ret == -ENOENT) {
1292 ntfs_debug("hiberfil.sys not present. Windows is not "
1293 "hibernated on the volume.");
1296 /* A real error occurred. */
1297 ntfs_error(vol->sb, "Failed to find inode number for "
1301 /* We do not care for the type of match that was found. */
1303 /* Get the inode. */
1304 vi = ntfs_iget(vol->sb, MREF(mref));
1305 if (IS_ERR(vi) || is_bad_inode(vi)) {
1308 ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1309 return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1311 if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1312 ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx). "
1313 "Windows is hibernated on the volume. This "
1314 "is not the system volume.", i_size_read(vi));
1317 page = ntfs_map_page(vi->i_mapping, 0);
1319 ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1320 ret = PTR_ERR(page);
1323 kaddr = (u32*)page_address(page);
1324 if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
1325 ntfs_debug("Magic \"hibr\" found in hiberfil.sys. Windows is "
1326 "hibernated on the volume. This is the "
1330 kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1332 if (unlikely(*kaddr)) {
1333 ntfs_debug("hiberfil.sys is larger than 4kiB "
1334 "(0x%llx), does not contain the "
1335 "\"hibr\" magic, and does not have a "
1336 "zero header. Windows is hibernated "
1337 "on the volume. This is not the "
1338 "system volume.", i_size_read(vi));
1341 } while (++kaddr < kend);
1342 ntfs_debug("hiberfil.sys contains a zero header. Windows is not "
1343 "hibernated on the volume. This is the system "
1347 ntfs_unmap_page(page);
1354 * load_and_init_quota - load and setup the quota file for a volume if present
1355 * @vol: ntfs super block describing device whose quota file to load
1357 * Return 'true' on success or 'false' on error. If $Quota is not present, we
1358 * leave vol->quota_ino as NULL and return success.
1360 static bool load_and_init_quota(ntfs_volume *vol)
1363 struct inode *tmp_ino;
1364 ntfs_name *name = NULL;
1365 static const ntfschar Quota[7] = { cpu_to_le16('$'),
1366 cpu_to_le16('Q'), cpu_to_le16('u'),
1367 cpu_to_le16('o'), cpu_to_le16('t'),
1368 cpu_to_le16('a'), 0 };
1369 static ntfschar Q[3] = { cpu_to_le16('$'),
1370 cpu_to_le16('Q'), 0 };
1372 ntfs_debug("Entering.");
1374 * Find the inode number for the quota file by looking up the filename
1375 * $Quota in the extended system files directory $Extend.
1377 mutex_lock(&vol->extend_ino->i_mutex);
1378 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1380 mutex_unlock(&vol->extend_ino->i_mutex);
1381 if (IS_ERR_MREF(mref)) {
1383 * If the file does not exist, quotas are disabled and have
1384 * never been enabled on this volume, just return success.
1386 if (MREF_ERR(mref) == -ENOENT) {
1387 ntfs_debug("$Quota not present. Volume does not have "
1390 * No need to try to set quotas out of date if they are
1393 NVolSetQuotaOutOfDate(vol);
1396 /* A real error occurred. */
1397 ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1400 /* We do not care for the type of match that was found. */
1402 /* Get the inode. */
1403 tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1404 if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1405 if (!IS_ERR(tmp_ino))
1407 ntfs_error(vol->sb, "Failed to load $Quota.");
1410 vol->quota_ino = tmp_ino;
1411 /* Get the $Q index allocation attribute. */
1412 tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1413 if (IS_ERR(tmp_ino)) {
1414 ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1417 vol->quota_q_ino = tmp_ino;
1418 ntfs_debug("Done.");
1423 * load_and_init_usnjrnl - load and setup the transaction log if present
1424 * @vol: ntfs super block describing device whose usnjrnl file to load
1426 * Return 'true' on success or 'false' on error.
1428 * If $UsnJrnl is not present or in the process of being disabled, we set
1429 * NVolUsnJrnlStamped() and return success.
1431 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1432 * i.e. transaction logging has only just been enabled or the journal has been
1433 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1434 * and return success.
1436 static bool load_and_init_usnjrnl(ntfs_volume *vol)
1439 struct inode *tmp_ino;
1442 ntfs_name *name = NULL;
1444 static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
1445 cpu_to_le16('U'), cpu_to_le16('s'),
1446 cpu_to_le16('n'), cpu_to_le16('J'),
1447 cpu_to_le16('r'), cpu_to_le16('n'),
1448 cpu_to_le16('l'), 0 };
1449 static ntfschar Max[5] = { cpu_to_le16('$'),
1450 cpu_to_le16('M'), cpu_to_le16('a'),
1451 cpu_to_le16('x'), 0 };
1452 static ntfschar J[3] = { cpu_to_le16('$'),
1453 cpu_to_le16('J'), 0 };
1455 ntfs_debug("Entering.");
1457 * Find the inode number for the transaction log file by looking up the
1458 * filename $UsnJrnl in the extended system files directory $Extend.
1460 mutex_lock(&vol->extend_ino->i_mutex);
1461 mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1463 mutex_unlock(&vol->extend_ino->i_mutex);
1464 if (IS_ERR_MREF(mref)) {
1466 * If the file does not exist, transaction logging is disabled,
1467 * just return success.
1469 if (MREF_ERR(mref) == -ENOENT) {
1470 ntfs_debug("$UsnJrnl not present. Volume does not "
1471 "have transaction logging enabled.");
1474 * No need to try to stamp the transaction log if
1475 * transaction logging is not enabled.
1477 NVolSetUsnJrnlStamped(vol);
1480 /* A real error occurred. */
1481 ntfs_error(vol->sb, "Failed to find inode number for "
1485 /* We do not care for the type of match that was found. */
1487 /* Get the inode. */
1488 tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1489 if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1490 if (!IS_ERR(tmp_ino))
1492 ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1495 vol->usnjrnl_ino = tmp_ino;
1497 * If the transaction log is in the process of being deleted, we can
1500 if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1501 ntfs_debug("$UsnJrnl in the process of being disabled. "
1502 "Volume does not have transaction logging "
1506 /* Get the $DATA/$Max attribute. */
1507 tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1508 if (IS_ERR(tmp_ino)) {
1509 ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1513 vol->usnjrnl_max_ino = tmp_ino;
1514 if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1515 ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1516 "attribute (size is 0x%llx but should be at "
1517 "least 0x%zx bytes).", i_size_read(tmp_ino),
1518 sizeof(USN_HEADER));
1521 /* Get the $DATA/$J attribute. */
1522 tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1523 if (IS_ERR(tmp_ino)) {
1524 ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1528 vol->usnjrnl_j_ino = tmp_ino;
1529 /* Verify $J is non-resident and sparse. */
1530 tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1531 if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1532 ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1533 "and/or not sparse.");
1536 /* Read the USN_HEADER from $DATA/$Max. */
1537 page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1539 ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1543 uh = (USN_HEADER*)page_address(page);
1544 /* Sanity check the $Max. */
1545 if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1546 sle64_to_cpu(uh->maximum_size))) {
1547 ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1548 "maximum size (0x%llx). $UsnJrnl is corrupt.",
1549 (long long)sle64_to_cpu(uh->allocation_delta),
1550 (long long)sle64_to_cpu(uh->maximum_size));
1551 ntfs_unmap_page(page);
1555 * If the transaction log has been stamped and nothing has been written
1556 * to it since, we do not need to stamp it.
1558 if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1559 i_size_read(vol->usnjrnl_j_ino))) {
1560 if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1561 i_size_read(vol->usnjrnl_j_ino))) {
1562 ntfs_unmap_page(page);
1563 ntfs_debug("$UsnJrnl is enabled but nothing has been "
1564 "logged since it was last stamped. "
1565 "Treating this as if the volume does "
1566 "not have transaction logging "
1570 ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1571 "which is out of bounds (0x%llx). $UsnJrnl "
1573 (long long)sle64_to_cpu(uh->lowest_valid_usn),
1574 i_size_read(vol->usnjrnl_j_ino));
1575 ntfs_unmap_page(page);
1578 ntfs_unmap_page(page);
1579 ntfs_debug("Done.");
1584 * load_and_init_attrdef - load the attribute definitions table for a volume
1585 * @vol: ntfs super block describing device whose attrdef to load
1587 * Return 'true' on success or 'false' on error.
1589 static bool load_and_init_attrdef(ntfs_volume *vol)
1592 struct super_block *sb = vol->sb;
1595 pgoff_t index, max_index;
1598 ntfs_debug("Entering.");
1599 /* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1600 ino = ntfs_iget(sb, FILE_AttrDef);
1601 if (IS_ERR(ino) || is_bad_inode(ino)) {
1606 NInoSetSparseDisabled(NTFS_I(ino));
1607 /* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1608 i_size = i_size_read(ino);
1609 if (i_size <= 0 || i_size > 0x7fffffff)
1611 vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1615 max_index = i_size >> PAGE_CACHE_SHIFT;
1616 size = PAGE_CACHE_SIZE;
1617 while (index < max_index) {
1618 /* Read the attrdef table and copy it into the linear buffer. */
1619 read_partial_attrdef_page:
1620 page = ntfs_map_page(ino->i_mapping, index);
1622 goto free_iput_failed;
1623 memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1624 page_address(page), size);
1625 ntfs_unmap_page(page);
1627 if (size == PAGE_CACHE_SIZE) {
1628 size = i_size & ~PAGE_CACHE_MASK;
1630 goto read_partial_attrdef_page;
1632 vol->attrdef_size = i_size;
1633 ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1637 ntfs_free(vol->attrdef);
1638 vol->attrdef = NULL;
1642 ntfs_error(sb, "Failed to initialize attribute definition table.");
1646 #endif /* NTFS_RW */
1649 * load_and_init_upcase - load the upcase table for an ntfs volume
1650 * @vol: ntfs super block describing device whose upcase to load
1652 * Return 'true' on success or 'false' on error.
1654 static bool load_and_init_upcase(ntfs_volume *vol)
1657 struct super_block *sb = vol->sb;
1660 pgoff_t index, max_index;
1664 ntfs_debug("Entering.");
1665 /* Read upcase table and setup vol->upcase and vol->upcase_len. */
1666 ino = ntfs_iget(sb, FILE_UpCase);
1667 if (IS_ERR(ino) || is_bad_inode(ino)) {
1673 * The upcase size must not be above 64k Unicode characters, must not
1674 * be zero and must be a multiple of sizeof(ntfschar).
1676 i_size = i_size_read(ino);
1677 if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1678 i_size > 64ULL * 1024 * sizeof(ntfschar))
1679 goto iput_upcase_failed;
1680 vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1682 goto iput_upcase_failed;
1684 max_index = i_size >> PAGE_CACHE_SHIFT;
1685 size = PAGE_CACHE_SIZE;
1686 while (index < max_index) {
1687 /* Read the upcase table and copy it into the linear buffer. */
1688 read_partial_upcase_page:
1689 page = ntfs_map_page(ino->i_mapping, index);
1691 goto iput_upcase_failed;
1692 memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1693 page_address(page), size);
1694 ntfs_unmap_page(page);
1696 if (size == PAGE_CACHE_SIZE) {
1697 size = i_size & ~PAGE_CACHE_MASK;
1699 goto read_partial_upcase_page;
1701 vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1702 ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1703 i_size, 64 * 1024 * sizeof(ntfschar));
1705 mutex_lock(&ntfs_lock);
1706 if (!default_upcase) {
1707 ntfs_debug("Using volume specified $UpCase since default is "
1709 mutex_unlock(&ntfs_lock);
1712 max = default_upcase_len;
1713 if (max > vol->upcase_len)
1714 max = vol->upcase_len;
1715 for (i = 0; i < max; i++)
1716 if (vol->upcase[i] != default_upcase[i])
1719 ntfs_free(vol->upcase);
1720 vol->upcase = default_upcase;
1721 vol->upcase_len = max;
1722 ntfs_nr_upcase_users++;
1723 mutex_unlock(&ntfs_lock);
1724 ntfs_debug("Volume specified $UpCase matches default. Using "
1728 mutex_unlock(&ntfs_lock);
1729 ntfs_debug("Using volume specified $UpCase since it does not match "
1734 ntfs_free(vol->upcase);
1737 mutex_lock(&ntfs_lock);
1738 if (default_upcase) {
1739 vol->upcase = default_upcase;
1740 vol->upcase_len = default_upcase_len;
1741 ntfs_nr_upcase_users++;
1742 mutex_unlock(&ntfs_lock);
1743 ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1747 mutex_unlock(&ntfs_lock);
1748 ntfs_error(sb, "Failed to initialize upcase table.");
1753 * The lcn and mft bitmap inodes are NTFS-internal inodes with
1754 * their own special locking rules:
1756 static struct lock_class_key
1757 lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1758 mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1761 * load_system_files - open the system files using normal functions
1762 * @vol: ntfs super block describing device whose system files to load
1764 * Open the system files with normal access functions and complete setting up
1765 * the ntfs super block @vol.
1767 * Return 'true' on success or 'false' on error.
1769 static bool load_system_files(ntfs_volume *vol)
1771 struct super_block *sb = vol->sb;
1773 VOLUME_INFORMATION *vi;
1774 ntfs_attr_search_ctx *ctx;
1776 RESTART_PAGE_HEADER *rp;
1778 #endif /* NTFS_RW */
1780 ntfs_debug("Entering.");
1782 /* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1783 if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1784 static const char *es1 = "Failed to load $MFTMirr";
1785 static const char *es2 = "$MFTMirr does not match $MFT";
1786 static const char *es3 = ". Run ntfsfix and/or chkdsk.";
1788 /* If a read-write mount, convert it to a read-only mount. */
1789 if (!(sb->s_flags & MS_RDONLY)) {
1790 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1791 ON_ERRORS_CONTINUE))) {
1792 ntfs_error(sb, "%s and neither on_errors="
1793 "continue nor on_errors="
1794 "remount-ro was specified%s",
1795 !vol->mftmirr_ino ? es1 : es2,
1797 goto iput_mirr_err_out;
1799 sb->s_flags |= MS_RDONLY;
1800 ntfs_error(sb, "%s. Mounting read-only%s",
1801 !vol->mftmirr_ino ? es1 : es2, es3);
1803 ntfs_warning(sb, "%s. Will not be able to remount "
1805 !vol->mftmirr_ino ? es1 : es2, es3);
1806 /* This will prevent a read-write remount. */
1809 #endif /* NTFS_RW */
1810 /* Get mft bitmap attribute inode. */
1811 vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1812 if (IS_ERR(vol->mftbmp_ino)) {
1813 ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1814 goto iput_mirr_err_out;
1816 lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1817 &mftbmp_runlist_lock_key);
1818 lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1819 &mftbmp_mrec_lock_key);
1820 /* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1821 if (!load_and_init_upcase(vol))
1822 goto iput_mftbmp_err_out;
1825 * Read attribute definitions table and setup @vol->attrdef and
1826 * @vol->attrdef_size.
1828 if (!load_and_init_attrdef(vol))
1829 goto iput_upcase_err_out;
1830 #endif /* NTFS_RW */
1832 * Get the cluster allocation bitmap inode and verify the size, no
1833 * need for any locking at this stage as we are already running
1834 * exclusively as we are mount in progress task.
1836 vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1837 if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1838 if (!IS_ERR(vol->lcnbmp_ino))
1839 iput(vol->lcnbmp_ino);
1842 lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1843 &lcnbmp_runlist_lock_key);
1844 lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1845 &lcnbmp_mrec_lock_key);
1847 NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1848 if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1849 iput(vol->lcnbmp_ino);
1851 ntfs_error(sb, "Failed to load $Bitmap.");
1852 goto iput_attrdef_err_out;
1855 * Get the volume inode and setup our cache of the volume flags and
1858 vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1859 if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1860 if (!IS_ERR(vol->vol_ino))
1863 ntfs_error(sb, "Failed to load $Volume.");
1864 goto iput_lcnbmp_err_out;
1866 m = map_mft_record(NTFS_I(vol->vol_ino));
1872 if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1873 ntfs_error(sb, "Failed to get attribute search context.");
1874 goto get_ctx_vol_failed;
1876 if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1877 ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1879 ntfs_attr_put_search_ctx(ctx);
1881 unmap_mft_record(NTFS_I(vol->vol_ino));
1882 goto iput_volume_failed;
1884 vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1885 le16_to_cpu(ctx->attr->data.resident.value_offset));
1886 /* Some bounds checks. */
1887 if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1888 le32_to_cpu(ctx->attr->data.resident.value_length) >
1889 (u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1891 /* Copy the volume flags and version to the ntfs_volume structure. */
1892 vol->vol_flags = vi->flags;
1893 vol->major_ver = vi->major_ver;
1894 vol->minor_ver = vi->minor_ver;
1895 ntfs_attr_put_search_ctx(ctx);
1896 unmap_mft_record(NTFS_I(vol->vol_ino));
1897 printk(KERN_INFO "NTFS volume version %i.%i.\n", vol->major_ver,
1899 if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1900 ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1901 "volume version %i.%i (need at least version "
1902 "3.0).", vol->major_ver, vol->minor_ver);
1903 NVolClearSparseEnabled(vol);
1906 /* Make sure that no unsupported volume flags are set. */
1907 if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1908 static const char *es1a = "Volume is dirty";
1909 static const char *es1b = "Volume has been modified by chkdsk";
1910 static const char *es1c = "Volume has unsupported flags set";
1911 static const char *es2a = ". Run chkdsk and mount in Windows.";
1912 static const char *es2b = ". Mount in Windows.";
1913 const char *es1, *es2;
1916 if (vol->vol_flags & VOLUME_IS_DIRTY)
1918 else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1923 ntfs_warning(sb, "Unsupported volume flags 0x%x "
1925 (unsigned)le16_to_cpu(vol->vol_flags));
1927 /* If a read-write mount, convert it to a read-only mount. */
1928 if (!(sb->s_flags & MS_RDONLY)) {
1929 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1930 ON_ERRORS_CONTINUE))) {
1931 ntfs_error(sb, "%s and neither on_errors="
1932 "continue nor on_errors="
1933 "remount-ro was specified%s",
1935 goto iput_vol_err_out;
1937 sb->s_flags |= MS_RDONLY;
1938 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1940 ntfs_warning(sb, "%s. Will not be able to remount "
1941 "read-write%s", es1, es2);
1943 * Do not set NVolErrors() because ntfs_remount() re-checks the
1944 * flags which we need to do in case any flags have changed.
1948 * Get the inode for the logfile, check it and determine if the volume
1949 * was shutdown cleanly.
1952 if (!load_and_check_logfile(vol, &rp) ||
1953 !ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1954 static const char *es1a = "Failed to load $LogFile";
1955 static const char *es1b = "$LogFile is not clean";
1956 static const char *es2 = ". Mount in Windows.";
1959 es1 = !vol->logfile_ino ? es1a : es1b;
1960 /* If a read-write mount, convert it to a read-only mount. */
1961 if (!(sb->s_flags & MS_RDONLY)) {
1962 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1963 ON_ERRORS_CONTINUE))) {
1964 ntfs_error(sb, "%s and neither on_errors="
1965 "continue nor on_errors="
1966 "remount-ro was specified%s",
1968 if (vol->logfile_ino) {
1972 goto iput_logfile_err_out;
1974 sb->s_flags |= MS_RDONLY;
1975 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1977 ntfs_warning(sb, "%s. Will not be able to remount "
1978 "read-write%s", es1, es2);
1979 /* This will prevent a read-write remount. */
1983 #endif /* NTFS_RW */
1984 /* Get the root directory inode so we can do path lookups. */
1985 vol->root_ino = ntfs_iget(sb, FILE_root);
1986 if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1987 if (!IS_ERR(vol->root_ino))
1988 iput(vol->root_ino);
1989 ntfs_error(sb, "Failed to load root directory.");
1990 goto iput_logfile_err_out;
1994 * Check if Windows is suspended to disk on the target volume. If it
1995 * is hibernated, we must not write *anything* to the disk so set
1996 * NVolErrors() without setting the dirty volume flag and mount
1997 * read-only. This will prevent read-write remounting and it will also
1998 * prevent all writes.
2000 err = check_windows_hibernation_status(vol);
2001 if (unlikely(err)) {
2002 static const char *es1a = "Failed to determine if Windows is "
2004 static const char *es1b = "Windows is hibernated";
2005 static const char *es2 = ". Run chkdsk.";
2008 es1 = err < 0 ? es1a : es1b;
2009 /* If a read-write mount, convert it to a read-only mount. */
2010 if (!(sb->s_flags & MS_RDONLY)) {
2011 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2012 ON_ERRORS_CONTINUE))) {
2013 ntfs_error(sb, "%s and neither on_errors="
2014 "continue nor on_errors="
2015 "remount-ro was specified%s",
2017 goto iput_root_err_out;
2019 sb->s_flags |= MS_RDONLY;
2020 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2022 ntfs_warning(sb, "%s. Will not be able to remount "
2023 "read-write%s", es1, es2);
2024 /* This will prevent a read-write remount. */
2027 /* If (still) a read-write mount, mark the volume dirty. */
2028 if (!(sb->s_flags & MS_RDONLY) &&
2029 ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
2030 static const char *es1 = "Failed to set dirty bit in volume "
2031 "information flags";
2032 static const char *es2 = ". Run chkdsk.";
2034 /* Convert to a read-only mount. */
2035 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2036 ON_ERRORS_CONTINUE))) {
2037 ntfs_error(sb, "%s and neither on_errors=continue nor "
2038 "on_errors=remount-ro was specified%s",
2040 goto iput_root_err_out;
2042 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2043 sb->s_flags |= MS_RDONLY;
2045 * Do not set NVolErrors() because ntfs_remount() might manage
2046 * to set the dirty flag in which case all would be well.
2050 // TODO: Enable this code once we start modifying anything that is
2051 // different between NTFS 1.2 and 3.x...
2053 * If (still) a read-write mount, set the NT4 compatibility flag on
2054 * newer NTFS version volumes.
2056 if (!(sb->s_flags & MS_RDONLY) && (vol->major_ver > 1) &&
2057 ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
2058 static const char *es1 = "Failed to set NT4 compatibility flag";
2059 static const char *es2 = ". Run chkdsk.";
2061 /* Convert to a read-only mount. */
2062 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2063 ON_ERRORS_CONTINUE))) {
2064 ntfs_error(sb, "%s and neither on_errors=continue nor "
2065 "on_errors=remount-ro was specified%s",
2067 goto iput_root_err_out;
2069 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2070 sb->s_flags |= MS_RDONLY;
2074 /* If (still) a read-write mount, empty the logfile. */
2075 if (!(sb->s_flags & MS_RDONLY) &&
2076 !ntfs_empty_logfile(vol->logfile_ino)) {
2077 static const char *es1 = "Failed to empty $LogFile";
2078 static const char *es2 = ". Mount in Windows.";
2080 /* Convert to a read-only mount. */
2081 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2082 ON_ERRORS_CONTINUE))) {
2083 ntfs_error(sb, "%s and neither on_errors=continue nor "
2084 "on_errors=remount-ro was specified%s",
2086 goto iput_root_err_out;
2088 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2089 sb->s_flags |= MS_RDONLY;
2092 #endif /* NTFS_RW */
2093 /* If on NTFS versions before 3.0, we are done. */
2094 if (unlikely(vol->major_ver < 3))
2096 /* NTFS 3.0+ specific initialization. */
2097 /* Get the security descriptors inode. */
2098 vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2099 if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2100 if (!IS_ERR(vol->secure_ino))
2101 iput(vol->secure_ino);
2102 ntfs_error(sb, "Failed to load $Secure.");
2103 goto iput_root_err_out;
2105 // TODO: Initialize security.
2106 /* Get the extended system files' directory inode. */
2107 vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2108 if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2109 if (!IS_ERR(vol->extend_ino))
2110 iput(vol->extend_ino);
2111 ntfs_error(sb, "Failed to load $Extend.");
2112 goto iput_sec_err_out;
2115 /* Find the quota file, load it if present, and set it up. */
2116 if (!load_and_init_quota(vol)) {
2117 static const char *es1 = "Failed to load $Quota";
2118 static const char *es2 = ". Run chkdsk.";
2120 /* If a read-write mount, convert it to a read-only mount. */
2121 if (!(sb->s_flags & MS_RDONLY)) {
2122 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2123 ON_ERRORS_CONTINUE))) {
2124 ntfs_error(sb, "%s and neither on_errors="
2125 "continue nor on_errors="
2126 "remount-ro was specified%s",
2128 goto iput_quota_err_out;
2130 sb->s_flags |= MS_RDONLY;
2131 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2133 ntfs_warning(sb, "%s. Will not be able to remount "
2134 "read-write%s", es1, es2);
2135 /* This will prevent a read-write remount. */
2138 /* If (still) a read-write mount, mark the quotas out of date. */
2139 if (!(sb->s_flags & MS_RDONLY) &&
2140 !ntfs_mark_quotas_out_of_date(vol)) {
2141 static const char *es1 = "Failed to mark quotas out of date";
2142 static const char *es2 = ". Run chkdsk.";
2144 /* Convert to a read-only mount. */
2145 if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2146 ON_ERRORS_CONTINUE))) {
2147 ntfs_error(sb, "%s and neither on_errors=continue nor "
2148 "on_errors=remount-ro was specified%s",
2150 goto iput_quota_err_out;
2152 ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2153 sb->s_flags |= MS_RDONLY;
2157 * Find the transaction log file ($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->s_flags & MS_RDONLY)) {
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 |= MS_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->s_flags & MS_RDONLY) && !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 |= MS_RDONLY;
2200 #endif /* NTFS_RW */
2203 iput_usnjrnl_err_out:
2204 if (vol->usnjrnl_j_ino)
2205 iput(vol->usnjrnl_j_ino);
2206 if (vol->usnjrnl_max_ino)
2207 iput(vol->usnjrnl_max_ino);
2208 if (vol->usnjrnl_ino)
2209 iput(vol->usnjrnl_ino);
2211 if (vol->quota_q_ino)
2212 iput(vol->quota_q_ino);
2214 iput(vol->quota_ino);
2215 iput(vol->extend_ino);
2216 #endif /* NTFS_RW */
2218 iput(vol->secure_ino);
2220 iput(vol->root_ino);
2221 iput_logfile_err_out:
2223 if (vol->logfile_ino)
2224 iput(vol->logfile_ino);
2226 #endif /* NTFS_RW */
2228 iput_lcnbmp_err_out:
2229 iput(vol->lcnbmp_ino);
2230 iput_attrdef_err_out:
2231 vol->attrdef_size = 0;
2233 ntfs_free(vol->attrdef);
2234 vol->attrdef = NULL;
2237 iput_upcase_err_out:
2238 #endif /* NTFS_RW */
2239 vol->upcase_len = 0;
2240 mutex_lock(&ntfs_lock);
2241 if (vol->upcase == default_upcase) {
2242 ntfs_nr_upcase_users--;
2245 mutex_unlock(&ntfs_lock);
2247 ntfs_free(vol->upcase);
2250 iput_mftbmp_err_out:
2251 iput(vol->mftbmp_ino);
2254 if (vol->mftmirr_ino)
2255 iput(vol->mftmirr_ino);
2256 #endif /* NTFS_RW */
2261 * ntfs_put_super - called by the vfs to unmount a volume
2262 * @sb: vfs superblock of volume to unmount
2264 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2265 * the volume is being unmounted (umount system call has been invoked) and it
2266 * releases all inodes and memory belonging to the NTFS specific part of the
2269 static void ntfs_put_super(struct super_block *sb)
2271 ntfs_volume *vol = NTFS_SB(sb);
2273 ntfs_debug("Entering.");
2277 * Commit all inodes while they are still open in case some of them
2278 * cause others to be dirtied.
2280 ntfs_commit_inode(vol->vol_ino);
2282 /* NTFS 3.0+ specific. */
2283 if (vol->major_ver >= 3) {
2284 if (vol->usnjrnl_j_ino)
2285 ntfs_commit_inode(vol->usnjrnl_j_ino);
2286 if (vol->usnjrnl_max_ino)
2287 ntfs_commit_inode(vol->usnjrnl_max_ino);
2288 if (vol->usnjrnl_ino)
2289 ntfs_commit_inode(vol->usnjrnl_ino);
2290 if (vol->quota_q_ino)
2291 ntfs_commit_inode(vol->quota_q_ino);
2293 ntfs_commit_inode(vol->quota_ino);
2294 if (vol->extend_ino)
2295 ntfs_commit_inode(vol->extend_ino);
2296 if (vol->secure_ino)
2297 ntfs_commit_inode(vol->secure_ino);
2300 ntfs_commit_inode(vol->root_ino);
2302 down_write(&vol->lcnbmp_lock);
2303 ntfs_commit_inode(vol->lcnbmp_ino);
2304 up_write(&vol->lcnbmp_lock);
2306 down_write(&vol->mftbmp_lock);
2307 ntfs_commit_inode(vol->mftbmp_ino);
2308 up_write(&vol->mftbmp_lock);
2310 if (vol->logfile_ino)
2311 ntfs_commit_inode(vol->logfile_ino);
2313 if (vol->mftmirr_ino)
2314 ntfs_commit_inode(vol->mftmirr_ino);
2315 ntfs_commit_inode(vol->mft_ino);
2318 * If a read-write mount and no volume errors have occurred, mark the
2319 * volume clean. Also, re-commit all affected inodes.
2321 if (!(sb->s_flags & MS_RDONLY)) {
2322 if (!NVolErrors(vol)) {
2323 if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2324 ntfs_warning(sb, "Failed to clear dirty bit "
2325 "in volume information "
2326 "flags. Run chkdsk.");
2327 ntfs_commit_inode(vol->vol_ino);
2328 ntfs_commit_inode(vol->root_ino);
2329 if (vol->mftmirr_ino)
2330 ntfs_commit_inode(vol->mftmirr_ino);
2331 ntfs_commit_inode(vol->mft_ino);
2333 ntfs_warning(sb, "Volume has errors. Leaving volume "
2334 "marked dirty. Run chkdsk.");
2337 #endif /* NTFS_RW */
2340 vol->vol_ino = NULL;
2342 /* NTFS 3.0+ specific clean up. */
2343 if (vol->major_ver >= 3) {
2345 if (vol->usnjrnl_j_ino) {
2346 iput(vol->usnjrnl_j_ino);
2347 vol->usnjrnl_j_ino = NULL;
2349 if (vol->usnjrnl_max_ino) {
2350 iput(vol->usnjrnl_max_ino);
2351 vol->usnjrnl_max_ino = NULL;
2353 if (vol->usnjrnl_ino) {
2354 iput(vol->usnjrnl_ino);
2355 vol->usnjrnl_ino = NULL;
2357 if (vol->quota_q_ino) {
2358 iput(vol->quota_q_ino);
2359 vol->quota_q_ino = NULL;
2361 if (vol->quota_ino) {
2362 iput(vol->quota_ino);
2363 vol->quota_ino = NULL;
2365 #endif /* NTFS_RW */
2366 if (vol->extend_ino) {
2367 iput(vol->extend_ino);
2368 vol->extend_ino = NULL;
2370 if (vol->secure_ino) {
2371 iput(vol->secure_ino);
2372 vol->secure_ino = NULL;
2376 iput(vol->root_ino);
2377 vol->root_ino = NULL;
2379 down_write(&vol->lcnbmp_lock);
2380 iput(vol->lcnbmp_ino);
2381 vol->lcnbmp_ino = NULL;
2382 up_write(&vol->lcnbmp_lock);
2384 down_write(&vol->mftbmp_lock);
2385 iput(vol->mftbmp_ino);
2386 vol->mftbmp_ino = NULL;
2387 up_write(&vol->mftbmp_lock);
2390 if (vol->logfile_ino) {
2391 iput(vol->logfile_ino);
2392 vol->logfile_ino = NULL;
2394 if (vol->mftmirr_ino) {
2395 /* Re-commit the mft mirror and mft just in case. */
2396 ntfs_commit_inode(vol->mftmirr_ino);
2397 ntfs_commit_inode(vol->mft_ino);
2398 iput(vol->mftmirr_ino);
2399 vol->mftmirr_ino = NULL;
2402 * We should have no dirty inodes left, due to
2403 * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2404 * the underlying mft records are written out and cleaned.
2406 ntfs_commit_inode(vol->mft_ino);
2407 write_inode_now(vol->mft_ino, 1);
2408 #endif /* NTFS_RW */
2411 vol->mft_ino = NULL;
2413 /* Throw away the table of attribute definitions. */
2414 vol->attrdef_size = 0;
2416 ntfs_free(vol->attrdef);
2417 vol->attrdef = NULL;
2419 vol->upcase_len = 0;
2421 * Destroy the global default upcase table if necessary. Also decrease
2422 * the number of upcase users if we are a user.
2424 mutex_lock(&ntfs_lock);
2425 if (vol->upcase == default_upcase) {
2426 ntfs_nr_upcase_users--;
2429 if (!ntfs_nr_upcase_users && default_upcase) {
2430 ntfs_free(default_upcase);
2431 default_upcase = NULL;
2433 if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2434 free_compression_buffers();
2435 mutex_unlock(&ntfs_lock);
2437 ntfs_free(vol->upcase);
2441 unload_nls(vol->nls_map);
2443 sb->s_fs_info = NULL;
2448 * get_nr_free_clusters - return the number of free clusters on a volume
2449 * @vol: ntfs volume for which to obtain free cluster count
2451 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2452 * actually calculate the number of clusters in use instead because this
2453 * allows us to not care about partial pages as these will be just zero filled
2454 * and hence not be counted as allocated clusters.
2456 * The only particularity is that clusters beyond the end of the logical ntfs
2457 * volume will be marked as allocated to prevent errors which means we have to
2458 * discount those at the end. This is important as the cluster bitmap always
2459 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2460 * the logical volume and marked in use when they are not as they do not exist.
2462 * If any pages cannot be read we assume all clusters in the erroring pages are
2463 * in use. This means we return an underestimate on errors which is better than
2466 static s64 get_nr_free_clusters(ntfs_volume *vol)
2468 s64 nr_free = vol->nr_clusters;
2469 struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2471 pgoff_t index, max_index;
2473 ntfs_debug("Entering.");
2474 /* Serialize accesses to the cluster bitmap. */
2475 down_read(&vol->lcnbmp_lock);
2477 * Convert the number of bits into bytes rounded up, then convert into
2478 * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2479 * full and one partial page max_index = 2.
2481 max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2483 /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2484 ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2485 max_index, PAGE_CACHE_SIZE / 4);
2486 for (index = 0; index < max_index; index++) {
2487 unsigned long *kaddr;
2490 * Read the page from page cache, getting it from backing store
2491 * if necessary, and increment the use count.
2493 page = read_mapping_page(mapping, index, NULL);
2494 /* Ignore pages which errored synchronously. */
2496 ntfs_debug("read_mapping_page() error. Skipping "
2497 "page (index 0x%lx).", index);
2498 nr_free -= PAGE_CACHE_SIZE * 8;
2501 kaddr = kmap_atomic(page);
2503 * Subtract the number of set bits. If this
2504 * is the last page and it is partial we don't really care as
2505 * it just means we do a little extra work but it won't affect
2506 * the result as all out of range bytes are set to zero by
2509 nr_free -= bitmap_weight(kaddr,
2510 PAGE_CACHE_SIZE * BITS_PER_BYTE);
2511 kunmap_atomic(kaddr);
2512 page_cache_release(page);
2514 ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2516 * Fixup for eventual bits outside logical ntfs volume (see function
2517 * description above).
2519 if (vol->nr_clusters & 63)
2520 nr_free += 64 - (vol->nr_clusters & 63);
2521 up_read(&vol->lcnbmp_lock);
2522 /* If errors occurred we may well have gone below zero, fix this. */
2525 ntfs_debug("Exiting.");
2530 * __get_nr_free_mft_records - return the number of free inodes on a volume
2531 * @vol: ntfs volume for which to obtain free inode count
2532 * @nr_free: number of mft records in filesystem
2533 * @max_index: maximum number of pages containing set bits
2535 * Calculate the number of free mft records (inodes) on the mounted NTFS
2536 * volume @vol. We actually calculate the number of mft records in use instead
2537 * because this allows us to not care about partial pages as these will be just
2538 * zero filled and hence not be counted as allocated mft record.
2540 * If any pages cannot be read we assume all mft records in the erroring pages
2541 * are in use. This means we return an underestimate on errors which is better
2542 * than an overestimate.
2544 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2546 static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2547 s64 nr_free, const pgoff_t max_index)
2549 struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2553 ntfs_debug("Entering.");
2554 /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2555 ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2556 "0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2557 for (index = 0; index < max_index; index++) {
2558 unsigned long *kaddr;
2561 * Read the page from page cache, getting it from backing store
2562 * if necessary, and increment the use count.
2564 page = read_mapping_page(mapping, index, NULL);
2565 /* Ignore pages which errored synchronously. */
2567 ntfs_debug("read_mapping_page() error. Skipping "
2568 "page (index 0x%lx).", index);
2569 nr_free -= PAGE_CACHE_SIZE * 8;
2572 kaddr = kmap_atomic(page);
2574 * Subtract the number of set bits. If this
2575 * is the last page and it is partial we don't really care as
2576 * it just means we do a little extra work but it won't affect
2577 * the result as all out of range bytes are set to zero by
2580 nr_free -= bitmap_weight(kaddr,
2581 PAGE_CACHE_SIZE * BITS_PER_BYTE);
2582 kunmap_atomic(kaddr);
2583 page_cache_release(page);
2585 ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2587 /* If errors occurred we may well have gone below zero, fix this. */
2590 ntfs_debug("Exiting.");
2595 * ntfs_statfs - return information about mounted NTFS volume
2596 * @dentry: dentry from mounted volume
2597 * @sfs: statfs structure in which to return the information
2599 * Return information about the mounted NTFS volume @dentry in the statfs structure
2600 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2601 * called). We interpret the values to be correct of the moment in time at
2602 * which we are called. Most values are variable otherwise and this isn't just
2603 * the free values but the totals as well. For example we can increase the
2604 * total number of file nodes if we run out and we can keep doing this until
2605 * there is no more space on the volume left at all.
2607 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2608 * ustat system calls.
2610 * Return 0 on success or -errno on error.
2612 static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2614 struct super_block *sb = dentry->d_sb;
2616 ntfs_volume *vol = NTFS_SB(sb);
2617 ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2619 unsigned long flags;
2621 ntfs_debug("Entering.");
2622 /* Type of filesystem. */
2623 sfs->f_type = NTFS_SB_MAGIC;
2624 /* Optimal transfer block size. */
2625 sfs->f_bsize = PAGE_CACHE_SIZE;
2627 * Total data blocks in filesystem in units of f_bsize and since
2628 * inodes are also stored in data blocs ($MFT is a file) this is just
2629 * the total clusters.
2631 sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2633 /* Free data blocks in filesystem in units of f_bsize. */
2634 size = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2638 /* Free blocks avail to non-superuser, same as above on NTFS. */
2639 sfs->f_bavail = sfs->f_bfree = size;
2640 /* Serialize accesses to the inode bitmap. */
2641 down_read(&vol->mftbmp_lock);
2642 read_lock_irqsave(&mft_ni->size_lock, flags);
2643 size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2645 * Convert the maximum number of set bits into bytes rounded up, then
2646 * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2647 * have one full and one partial page max_index = 2.
2649 max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2650 + 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2651 read_unlock_irqrestore(&mft_ni->size_lock, flags);
2652 /* Number of inodes in filesystem (at this point in time). */
2653 sfs->f_files = size;
2654 /* Free inodes in fs (based on current total count). */
2655 sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2656 up_read(&vol->mftbmp_lock);
2658 * File system id. This is extremely *nix flavour dependent and even
2659 * within Linux itself all fs do their own thing. I interpret this to
2660 * mean a unique id associated with the mounted fs and not the id
2661 * associated with the filesystem driver, the latter is already given
2662 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2663 * volume serial number splitting it into two 32-bit parts. We enter
2664 * the least significant 32-bits in f_fsid[0] and the most significant
2665 * 32-bits in f_fsid[1].
2667 sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2668 sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2669 /* Maximum length of filenames. */
2670 sfs->f_namelen = NTFS_MAX_NAME_LEN;
2675 static int ntfs_write_inode(struct inode *vi, struct writeback_control *wbc)
2677 return __ntfs_write_inode(vi, wbc->sync_mode == WB_SYNC_ALL);
2682 * The complete super operations.
2684 static const struct super_operations ntfs_sops = {
2685 .alloc_inode = ntfs_alloc_big_inode, /* VFS: Allocate new inode. */
2686 .destroy_inode = ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2688 .write_inode = ntfs_write_inode, /* VFS: Write dirty inode to
2690 #endif /* NTFS_RW */
2691 .put_super = ntfs_put_super, /* Syscall: umount. */
2692 .statfs = ntfs_statfs, /* Syscall: statfs */
2693 .remount_fs = ntfs_remount, /* Syscall: mount -o remount. */
2694 .evict_inode = ntfs_evict_big_inode, /* VFS: Called when an inode is
2695 removed from memory. */
2696 .show_options = ntfs_show_options, /* Show mount options in
2701 * ntfs_fill_super - mount an ntfs filesystem
2702 * @sb: super block of ntfs filesystem to mount
2703 * @opt: string containing the mount options
2704 * @silent: silence error output
2706 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2707 * with the mount otions in @data with the NTFS filesystem.
2709 * If @silent is true, remain silent even if errors are detected. This is used
2710 * during bootup, when the kernel tries to mount the root filesystem with all
2711 * registered filesystems one after the other until one succeeds. This implies
2712 * that all filesystems except the correct one will quite correctly and
2713 * expectedly return an error, but nobody wants to see error messages when in
2714 * fact this is what is supposed to happen.
2716 * NOTE: @sb->s_flags contains the mount options flags.
2718 static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2721 struct buffer_head *bh;
2722 struct inode *tmp_ino;
2723 int blocksize, result;
2726 * We do a pretty difficult piece of bootstrap by reading the
2727 * MFT (and other metadata) from disk into memory. We'll only
2728 * release this metadata during umount, so the locking patterns
2729 * observed during bootstrap do not count. So turn off the
2730 * observation of locking patterns (strictly for this context
2731 * only) while mounting NTFS. [The validator is still active
2732 * otherwise, even for this context: it will for example record
2733 * lock class registrations.]
2736 ntfs_debug("Entering.");
2738 sb->s_flags |= MS_RDONLY;
2739 #endif /* ! NTFS_RW */
2740 /* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2741 sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2745 ntfs_error(sb, "Allocation of NTFS volume structure "
2746 "failed. Aborting mount...");
2750 /* Initialize ntfs_volume structure. */
2751 *vol = (ntfs_volume) {
2754 * Default is group and other don't have any access to files or
2755 * directories while owner has full access. Further, files by
2756 * default are not executable but directories are of course
2762 init_rwsem(&vol->mftbmp_lock);
2763 init_rwsem(&vol->lcnbmp_lock);
2765 /* By default, enable sparse support. */
2766 NVolSetSparseEnabled(vol);
2768 /* Important to get the mount options dealt with now. */
2769 if (!parse_options(vol, (char*)opt))
2772 /* We support sector sizes up to the PAGE_CACHE_SIZE. */
2773 if (bdev_logical_block_size(sb->s_bdev) > PAGE_CACHE_SIZE) {
2775 ntfs_error(sb, "Device has unsupported sector size "
2776 "(%i). The maximum supported sector "
2777 "size on this architecture is %lu "
2779 bdev_logical_block_size(sb->s_bdev),
2784 * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2785 * sector size, whichever is bigger.
2787 blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2788 if (blocksize < NTFS_BLOCK_SIZE) {
2790 ntfs_error(sb, "Unable to set device block size.");
2793 BUG_ON(blocksize != sb->s_blocksize);
2794 ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2795 blocksize, sb->s_blocksize_bits);
2796 /* Determine the size of the device in units of block_size bytes. */
2797 if (!i_size_read(sb->s_bdev->bd_inode)) {
2799 ntfs_error(sb, "Unable to determine device size.");
2802 vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2803 sb->s_blocksize_bits;
2804 /* Read the boot sector and return unlocked buffer head to it. */
2805 if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2807 ntfs_error(sb, "Not an NTFS volume.");
2811 * Extract the data from the boot sector and setup the ntfs volume
2814 result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2818 ntfs_error(sb, "Unsupported NTFS filesystem.");
2822 * If the boot sector indicates a sector size bigger than the current
2823 * device block size, switch the device block size to the sector size.
2824 * TODO: It may be possible to support this case even when the set
2825 * below fails, we would just be breaking up the i/o for each sector
2826 * into multiple blocks for i/o purposes but otherwise it should just
2827 * work. However it is safer to leave disabled until someone hits this
2828 * error message and then we can get them to try it without the setting
2829 * so we know for sure that it works.
2831 if (vol->sector_size > blocksize) {
2832 blocksize = sb_set_blocksize(sb, vol->sector_size);
2833 if (blocksize != vol->sector_size) {
2835 ntfs_error(sb, "Unable to set device block "
2836 "size to sector size (%i).",
2840 BUG_ON(blocksize != sb->s_blocksize);
2841 vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2842 sb->s_blocksize_bits;
2843 ntfs_debug("Changed device block size to %i bytes (block size "
2844 "bits %i) to match volume sector size.",
2845 blocksize, sb->s_blocksize_bits);
2847 /* Initialize the cluster and mft allocators. */
2848 ntfs_setup_allocators(vol);
2849 /* Setup remaining fields in the super block. */
2850 sb->s_magic = NTFS_SB_MAGIC;
2852 * Ntfs allows 63 bits for the file size, i.e. correct would be:
2853 * sb->s_maxbytes = ~0ULL >> 1;
2854 * But the kernel uses a long as the page cache page index which on
2855 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2856 * defined to the maximum the page cache page index can cope with
2857 * without overflowing the index or to 2^63 - 1, whichever is smaller.
2859 sb->s_maxbytes = MAX_LFS_FILESIZE;
2860 /* Ntfs measures time in 100ns intervals. */
2861 sb->s_time_gran = 100;
2863 * Now load the metadata required for the page cache and our address
2864 * space operations to function. We do this by setting up a specialised
2865 * read_inode method and then just calling the normal iget() to obtain
2866 * the inode for $MFT which is sufficient to allow our normal inode
2867 * operations and associated address space operations to function.
2869 sb->s_op = &ntfs_sops;
2870 tmp_ino = new_inode(sb);
2873 ntfs_error(sb, "Failed to load essential metadata.");
2876 tmp_ino->i_ino = FILE_MFT;
2877 insert_inode_hash(tmp_ino);
2878 if (ntfs_read_inode_mount(tmp_ino) < 0) {
2880 ntfs_error(sb, "Failed to load essential metadata.");
2881 goto iput_tmp_ino_err_out_now;
2883 mutex_lock(&ntfs_lock);
2885 * The current mount is a compression user if the cluster size is
2886 * less than or equal 4kiB.
2888 if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2889 result = allocate_compression_buffers();
2891 ntfs_error(NULL, "Failed to allocate buffers "
2892 "for compression engine.");
2893 ntfs_nr_compression_users--;
2894 mutex_unlock(&ntfs_lock);
2895 goto iput_tmp_ino_err_out_now;
2899 * Generate the global default upcase table if necessary. Also
2900 * temporarily increment the number of upcase users to avoid race
2901 * conditions with concurrent (u)mounts.
2903 if (!default_upcase)
2904 default_upcase = generate_default_upcase();
2905 ntfs_nr_upcase_users++;
2906 mutex_unlock(&ntfs_lock);
2908 * From now on, ignore @silent parameter. If we fail below this line,
2909 * it will be due to a corrupt fs or a system error, so we report it.
2912 * Open the system files with normal access functions and complete
2913 * setting up the ntfs super block.
2915 if (!load_system_files(vol)) {
2916 ntfs_error(sb, "Failed to load system files.");
2917 goto unl_upcase_iput_tmp_ino_err_out_now;
2920 /* We grab a reference, simulating an ntfs_iget(). */
2921 ihold(vol->root_ino);
2922 if ((sb->s_root = d_make_root(vol->root_ino))) {
2923 ntfs_debug("Exiting, status successful.");
2924 /* Release the default upcase if it has no users. */
2925 mutex_lock(&ntfs_lock);
2926 if (!--ntfs_nr_upcase_users && default_upcase) {
2927 ntfs_free(default_upcase);
2928 default_upcase = NULL;
2930 mutex_unlock(&ntfs_lock);
2931 sb->s_export_op = &ntfs_export_ops;
2935 ntfs_error(sb, "Failed to allocate root directory.");
2936 /* Clean up after the successful load_system_files() call from above. */
2937 // TODO: Use ntfs_put_super() instead of repeating all this code...
2938 // FIXME: Should mark the volume clean as the error is most likely
2941 vol->vol_ino = NULL;
2942 /* NTFS 3.0+ specific clean up. */
2943 if (vol->major_ver >= 3) {
2945 if (vol->usnjrnl_j_ino) {
2946 iput(vol->usnjrnl_j_ino);
2947 vol->usnjrnl_j_ino = NULL;
2949 if (vol->usnjrnl_max_ino) {
2950 iput(vol->usnjrnl_max_ino);
2951 vol->usnjrnl_max_ino = NULL;
2953 if (vol->usnjrnl_ino) {
2954 iput(vol->usnjrnl_ino);
2955 vol->usnjrnl_ino = NULL;
2957 if (vol->quota_q_ino) {
2958 iput(vol->quota_q_ino);
2959 vol->quota_q_ino = NULL;
2961 if (vol->quota_ino) {
2962 iput(vol->quota_ino);
2963 vol->quota_ino = NULL;
2965 #endif /* NTFS_RW */
2966 if (vol->extend_ino) {
2967 iput(vol->extend_ino);
2968 vol->extend_ino = NULL;
2970 if (vol->secure_ino) {
2971 iput(vol->secure_ino);
2972 vol->secure_ino = NULL;
2975 iput(vol->root_ino);
2976 vol->root_ino = NULL;
2977 iput(vol->lcnbmp_ino);
2978 vol->lcnbmp_ino = NULL;
2979 iput(vol->mftbmp_ino);
2980 vol->mftbmp_ino = NULL;
2982 if (vol->logfile_ino) {
2983 iput(vol->logfile_ino);
2984 vol->logfile_ino = NULL;
2986 if (vol->mftmirr_ino) {
2987 iput(vol->mftmirr_ino);
2988 vol->mftmirr_ino = NULL;
2990 #endif /* NTFS_RW */
2991 /* Throw away the table of attribute definitions. */
2992 vol->attrdef_size = 0;
2994 ntfs_free(vol->attrdef);
2995 vol->attrdef = NULL;
2997 vol->upcase_len = 0;
2998 mutex_lock(&ntfs_lock);
2999 if (vol->upcase == default_upcase) {
3000 ntfs_nr_upcase_users--;
3003 mutex_unlock(&ntfs_lock);
3005 ntfs_free(vol->upcase);
3009 unload_nls(vol->nls_map);
3010 vol->nls_map = NULL;
3012 /* Error exit code path. */
3013 unl_upcase_iput_tmp_ino_err_out_now:
3015 * Decrease the number of upcase users and destroy the global default
3016 * upcase table if necessary.
3018 mutex_lock(&ntfs_lock);
3019 if (!--ntfs_nr_upcase_users && default_upcase) {
3020 ntfs_free(default_upcase);
3021 default_upcase = NULL;
3023 if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
3024 free_compression_buffers();
3025 mutex_unlock(&ntfs_lock);
3026 iput_tmp_ino_err_out_now:
3028 if (vol->mft_ino && vol->mft_ino != tmp_ino)
3030 vol->mft_ino = NULL;
3031 /* Errors at this stage are irrelevant. */
3033 sb->s_fs_info = NULL;
3035 ntfs_debug("Failed, returning -EINVAL.");
3041 * This is a slab cache to optimize allocations and deallocations of Unicode
3042 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3043 * (255) Unicode characters + a terminating NULL Unicode character.
3045 struct kmem_cache *ntfs_name_cache;
3047 /* Slab caches for efficient allocation/deallocation of inodes. */
3048 struct kmem_cache *ntfs_inode_cache;
3049 struct kmem_cache *ntfs_big_inode_cache;
3051 /* Init once constructor for the inode slab cache. */
3052 static void ntfs_big_inode_init_once(void *foo)
3054 ntfs_inode *ni = (ntfs_inode *)foo;
3056 inode_init_once(VFS_I(ni));
3060 * Slab caches to optimize allocations and deallocations of attribute search
3061 * contexts and index contexts, respectively.
3063 struct kmem_cache *ntfs_attr_ctx_cache;
3064 struct kmem_cache *ntfs_index_ctx_cache;
3066 /* Driver wide mutex. */
3067 DEFINE_MUTEX(ntfs_lock);
3069 static struct dentry *ntfs_mount(struct file_system_type *fs_type,
3070 int flags, const char *dev_name, void *data)
3072 return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
3075 static struct file_system_type ntfs_fs_type = {
3076 .owner = THIS_MODULE,
3078 .mount = ntfs_mount,
3079 .kill_sb = kill_block_super,
3080 .fs_flags = FS_REQUIRES_DEV,
3083 /* Stable names for the slab caches. */
3084 static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3085 static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3086 static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3087 static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3088 static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3090 static int __init init_ntfs_fs(void)
3094 /* This may be ugly but it results in pretty output so who cares. (-8 */
3095 printk(KERN_INFO "NTFS driver " NTFS_VERSION " [Flags: R/"
3109 ntfs_debug("Debug messages are enabled.");
3111 ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3112 sizeof(ntfs_index_context), 0 /* offset */,
3113 SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3114 if (!ntfs_index_ctx_cache) {
3115 printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3116 ntfs_index_ctx_cache_name);
3119 ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3120 sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3121 SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3122 if (!ntfs_attr_ctx_cache) {
3123 printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3124 ntfs_attr_ctx_cache_name);
3128 ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3129 (NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3130 SLAB_HWCACHE_ALIGN, NULL);
3131 if (!ntfs_name_cache) {
3132 printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3133 ntfs_name_cache_name);
3137 ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3138 sizeof(ntfs_inode), 0,
3139 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
3140 if (!ntfs_inode_cache) {
3141 printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3142 ntfs_inode_cache_name);
3146 ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3147 sizeof(big_ntfs_inode), 0,
3148 SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
3149 ntfs_big_inode_init_once);
3150 if (!ntfs_big_inode_cache) {
3151 printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3152 ntfs_big_inode_cache_name);
3153 goto big_inode_err_out;
3156 /* Register the ntfs sysctls. */
3157 err = ntfs_sysctl(1);
3159 printk(KERN_CRIT "NTFS: Failed to register NTFS sysctls!\n");
3160 goto sysctl_err_out;
3163 err = register_filesystem(&ntfs_fs_type);
3165 ntfs_debug("NTFS driver registered successfully.");
3166 return 0; /* Success! */
3168 printk(KERN_CRIT "NTFS: Failed to register NTFS filesystem driver!\n");
3170 /* Unregister the ntfs sysctls. */
3173 kmem_cache_destroy(ntfs_big_inode_cache);
3175 kmem_cache_destroy(ntfs_inode_cache);
3177 kmem_cache_destroy(ntfs_name_cache);
3179 kmem_cache_destroy(ntfs_attr_ctx_cache);
3181 kmem_cache_destroy(ntfs_index_ctx_cache);
3184 printk(KERN_CRIT "NTFS: Aborting NTFS filesystem driver "
3185 "registration...\n");
3191 static void __exit exit_ntfs_fs(void)
3193 ntfs_debug("Unregistering NTFS driver.");
3195 unregister_filesystem(&ntfs_fs_type);
3198 * Make sure all delayed rcu free inodes are flushed before we
3202 kmem_cache_destroy(ntfs_big_inode_cache);
3203 kmem_cache_destroy(ntfs_inode_cache);
3204 kmem_cache_destroy(ntfs_name_cache);
3205 kmem_cache_destroy(ntfs_attr_ctx_cache);
3206 kmem_cache_destroy(ntfs_index_ctx_cache);
3207 /* Unregister the ntfs sysctls. */
3211 MODULE_AUTHOR("Anton Altaparmakov <anton@tuxera.com>");
3212 MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2011 Anton Altaparmakov and Tuxera Inc.");
3213 MODULE_VERSION(NTFS_VERSION);
3214 MODULE_LICENSE("GPL");
3216 module_param(debug_msgs, bint, 0);
3217 MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3220 module_init(init_ntfs_fs)
3221 module_exit(exit_ntfs_fs)