Root/fs/ntfs/super.c

1/*
2 * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
3 *
4 * Copyright (c) 2001-2011 Anton Altaparmakov and Tuxera Inc.
5 * Copyright (c) 2001,2002 Richard Russon
6 *
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.
11 *
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.
16 *
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
21 */
22
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>
34
35#include "sysctl.h"
36#include "logfile.h"
37#include "quota.h"
38#include "usnjrnl.h"
39#include "dir.h"
40#include "debug.h"
41#include "index.h"
42#include "inode.h"
43#include "aops.h"
44#include "layout.h"
45#include "malloc.h"
46#include "ntfs.h"
47
48/* Number of mounted filesystems which have compression enabled. */
49static unsigned long ntfs_nr_compression_users;
50
51/* A global default upcase table and a corresponding reference count. */
52static ntfschar *default_upcase = NULL;
53static unsigned long ntfs_nr_upcase_users = 0;
54
55/* Error constants/strings used in inode.c::ntfs_show_options(). */
56typedef enum {
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,
63} ON_ERRORS_ACTIONS;
64
65const 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" },
70    { 0, NULL }
71};
72
73/**
74 * simple_getbool -
75 *
76 * Copied from old ntfs driver (which copied from vfat driver).
77 */
78static int simple_getbool(char *s, bool *setval)
79{
80    if (s) {
81        if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
82            *setval = true;
83        else if (!strcmp(s, "0") || !strcmp(s, "no") ||
84                            !strcmp(s, "false"))
85            *setval = false;
86        else
87            return 0;
88    } else
89        *setval = true;
90    return 1;
91}
92
93/**
94 * parse_options - parse the (re)mount options
95 * @vol: ntfs volume
96 * @opt: string containing the (re)mount options
97 *
98 * Parse the recognized options in @opt for the ntfs volume described by @vol.
99 */
100static bool parse_options(ntfs_volume *vol, char *opt)
101{
102    char *p, *v, *ov;
103    static char *utf8 = "utf8";
104    int errors = 0, sloppy = 0;
105    uid_t uid = (uid_t)-1;
106    gid_t gid = (gid_t)-1;
107    mode_t fmask = (mode_t)-1, dmask = (mode_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;
111
112    /* I am lazy... (-8 */
113#define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value) \
114    if (!strcmp(p, option)) { \
115        if (!v || !*v) \
116            variable = default_value; \
117        else { \
118            variable = simple_strtoul(ov = v, &v, 0); \
119            if (*v) \
120                goto needs_val; \
121        } \
122    }
123#define NTFS_GETOPT(option, variable) \
124    if (!strcmp(p, option)) { \
125        if (!v || !*v) \
126            goto needs_arg; \
127        variable = simple_strtoul(ov = v, &v, 0); \
128        if (*v) \
129            goto needs_val; \
130    }
131#define NTFS_GETOPT_OCTAL(option, variable) \
132    if (!strcmp(p, option)) { \
133        if (!v || !*v) \
134            goto needs_arg; \
135        variable = simple_strtoul(ov = v, &v, 8); \
136        if (*v) \
137            goto needs_val; \
138    }
139#define NTFS_GETOPT_BOOL(option, variable) \
140    if (!strcmp(p, option)) { \
141        bool val; \
142        if (!simple_getbool(v, &val)) \
143            goto needs_bool; \
144        variable = val; \
145    }
146#define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array) \
147    if (!strcmp(p, option)) { \
148        int _i; \
149        if (!v || !*v) \
150            goto needs_arg; \
151        ov = v; \
152        if (variable == -1) \
153            variable = 0; \
154        for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
155            if (!strcmp(opt_array[_i].str, v)) { \
156                variable |= opt_array[_i].val; \
157                break; \
158            } \
159        if (!opt_array[_i].str || !*opt_array[_i].str) \
160            goto needs_val; \
161    }
162    if (!opt || !*opt)
163        goto no_mount_options;
164    ntfs_debug("Entering with mount options string: %s", opt);
165    while ((p = strsep(&opt, ","))) {
166        if ((v = strchr(p, '=')))
167            *v++ = 0;
168        NTFS_GETOPT("uid", uid)
169        else NTFS_GETOPT("gid", gid)
170        else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
171        else NTFS_GETOPT_OCTAL("fmask", fmask)
172        else NTFS_GETOPT_OCTAL("dmask", dmask)
173        else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
174        else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
175        else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
176        else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
177        else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
178        else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
179                on_errors_arr)
180        else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
181            ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
182                    p);
183        else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
184            if (!strcmp(p, "iocharset"))
185                ntfs_warning(vol->sb, "Option iocharset is "
186                        "deprecated. Please use "
187                        "option nls=<charsetname> in "
188                        "the future.");
189            if (!v || !*v)
190                goto needs_arg;
191use_utf8:
192            old_nls = nls_map;
193            nls_map = load_nls(v);
194            if (!nls_map) {
195                if (!old_nls) {
196                    ntfs_error(vol->sb, "NLS character set "
197                            "%s not found.", v);
198                    return false;
199                }
200                ntfs_error(vol->sb, "NLS character set %s not "
201                        "found. Using previous one %s.",
202                        v, old_nls->charset);
203                nls_map = old_nls;
204            } else /* nls_map */ {
205                unload_nls(old_nls);
206            }
207        } else if (!strcmp(p, "utf8")) {
208            bool val = false;
209            ntfs_warning(vol->sb, "Option utf8 is no longer "
210                   "supported, using option nls=utf8. Please "
211                   "use option nls=utf8 in the future and "
212                   "make sure utf8 is compiled either as a "
213                   "module or into the kernel.");
214            if (!v || !*v)
215                val = true;
216            else if (!simple_getbool(v, &val))
217                goto needs_bool;
218            if (val) {
219                v = utf8;
220                goto use_utf8;
221            }
222        } else {
223            ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
224            if (errors < INT_MAX)
225                errors++;
226        }
227#undef NTFS_GETOPT_OPTIONS_ARRAY
228#undef NTFS_GETOPT_BOOL
229#undef NTFS_GETOPT
230#undef NTFS_GETOPT_WITH_DEFAULT
231    }
232no_mount_options:
233    if (errors && !sloppy)
234        return false;
235    if (sloppy)
236        ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
237                "unrecognized mount option(s) and continuing.");
238    /* Keep this first! */
239    if (on_errors != -1) {
240        if (!on_errors) {
241            ntfs_error(vol->sb, "Invalid errors option argument "
242                    "or bug in options parser.");
243            return false;
244        }
245    }
246    if (nls_map) {
247        if (vol->nls_map && vol->nls_map != nls_map) {
248            ntfs_error(vol->sb, "Cannot change NLS character set "
249                    "on remount.");
250            return false;
251        } /* else (!vol->nls_map) */
252        ntfs_debug("Using NLS character set %s.", nls_map->charset);
253        vol->nls_map = nls_map;
254    } else /* (!nls_map) */ {
255        if (!vol->nls_map) {
256            vol->nls_map = load_nls_default();
257            if (!vol->nls_map) {
258                ntfs_error(vol->sb, "Failed to load default "
259                        "NLS character set.");
260                return false;
261            }
262            ntfs_debug("Using default NLS character set (%s).",
263                    vol->nls_map->charset);
264        }
265    }
266    if (mft_zone_multiplier != -1) {
267        if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
268                mft_zone_multiplier) {
269            ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
270                    "on remount.");
271            return false;
272        }
273        if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
274            ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
275                    "Using default value, i.e. 1.");
276            mft_zone_multiplier = 1;
277        }
278        vol->mft_zone_multiplier = mft_zone_multiplier;
279    }
280    if (!vol->mft_zone_multiplier)
281        vol->mft_zone_multiplier = 1;
282    if (on_errors != -1)
283        vol->on_errors = on_errors;
284    if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
285        vol->on_errors |= ON_ERRORS_CONTINUE;
286    if (uid != (uid_t)-1)
287        vol->uid = uid;
288    if (gid != (gid_t)-1)
289        vol->gid = gid;
290    if (fmask != (mode_t)-1)
291        vol->fmask = fmask;
292    if (dmask != (mode_t)-1)
293        vol->dmask = dmask;
294    if (show_sys_files != -1) {
295        if (show_sys_files)
296            NVolSetShowSystemFiles(vol);
297        else
298            NVolClearShowSystemFiles(vol);
299    }
300    if (case_sensitive != -1) {
301        if (case_sensitive)
302            NVolSetCaseSensitive(vol);
303        else
304            NVolClearCaseSensitive(vol);
305    }
306    if (disable_sparse != -1) {
307        if (disable_sparse)
308            NVolClearSparseEnabled(vol);
309        else {
310            if (!NVolSparseEnabled(vol) &&
311                    vol->major_ver && vol->major_ver < 3)
312                ntfs_warning(vol->sb, "Not enabling sparse "
313                        "support due to NTFS volume "
314                        "version %i.%i (need at least "
315                        "version 3.0).", vol->major_ver,
316                        vol->minor_ver);
317            else
318                NVolSetSparseEnabled(vol);
319        }
320    }
321    return true;
322needs_arg:
323    ntfs_error(vol->sb, "The %s option requires an argument.", p);
324    return false;
325needs_bool:
326    ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
327    return false;
328needs_val:
329    ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
330    return false;
331}
332
333#ifdef NTFS_RW
334
335/**
336 * ntfs_write_volume_flags - write new flags to the volume information flags
337 * @vol: ntfs volume on which to modify the flags
338 * @flags: new flags value for the volume information flags
339 *
340 * Internal function. You probably want to use ntfs_{set,clear}_volume_flags()
341 * instead (see below).
342 *
343 * Replace the volume information flags on the volume @vol with the value
344 * supplied in @flags. Note, this overwrites the volume information flags, so
345 * make sure to combine the flags you want to modify with the old flags and use
346 * the result when calling ntfs_write_volume_flags().
347 *
348 * Return 0 on success and -errno on error.
349 */
350static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
351{
352    ntfs_inode *ni = NTFS_I(vol->vol_ino);
353    MFT_RECORD *m;
354    VOLUME_INFORMATION *vi;
355    ntfs_attr_search_ctx *ctx;
356    int err;
357
358    ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
359            le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
360    if (vol->vol_flags == flags)
361        goto done;
362    BUG_ON(!ni);
363    m = map_mft_record(ni);
364    if (IS_ERR(m)) {
365        err = PTR_ERR(m);
366        goto err_out;
367    }
368    ctx = ntfs_attr_get_search_ctx(ni, m);
369    if (!ctx) {
370        err = -ENOMEM;
371        goto put_unm_err_out;
372    }
373    err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
374            ctx);
375    if (err)
376        goto put_unm_err_out;
377    vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
378            le16_to_cpu(ctx->attr->data.resident.value_offset));
379    vol->vol_flags = vi->flags = flags;
380    flush_dcache_mft_record_page(ctx->ntfs_ino);
381    mark_mft_record_dirty(ctx->ntfs_ino);
382    ntfs_attr_put_search_ctx(ctx);
383    unmap_mft_record(ni);
384done:
385    ntfs_debug("Done.");
386    return 0;
387put_unm_err_out:
388    if (ctx)
389        ntfs_attr_put_search_ctx(ctx);
390    unmap_mft_record(ni);
391err_out:
392    ntfs_error(vol->sb, "Failed with error code %i.", -err);
393    return err;
394}
395
396/**
397 * ntfs_set_volume_flags - set bits in the volume information flags
398 * @vol: ntfs volume on which to modify the flags
399 * @flags: flags to set on the volume
400 *
401 * Set the bits in @flags in the volume information flags on the volume @vol.
402 *
403 * Return 0 on success and -errno on error.
404 */
405static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
406{
407    flags &= VOLUME_FLAGS_MASK;
408    return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
409}
410
411/**
412 * ntfs_clear_volume_flags - clear bits in the volume information flags
413 * @vol: ntfs volume on which to modify the flags
414 * @flags: flags to clear on the volume
415 *
416 * Clear the bits in @flags in the volume information flags on the volume @vol.
417 *
418 * Return 0 on success and -errno on error.
419 */
420static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
421{
422    flags &= VOLUME_FLAGS_MASK;
423    flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
424    return ntfs_write_volume_flags(vol, flags);
425}
426
427#endif /* NTFS_RW */
428
429/**
430 * ntfs_remount - change the mount options of a mounted ntfs filesystem
431 * @sb: superblock of mounted ntfs filesystem
432 * @flags: remount flags
433 * @opt: remount options string
434 *
435 * Change the mount options of an already mounted ntfs filesystem.
436 *
437 * NOTE: The VFS sets the @sb->s_flags remount flags to @flags after
438 * ntfs_remount() returns successfully (i.e. returns 0). Otherwise,
439 * @sb->s_flags are not changed.
440 */
441static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
442{
443    ntfs_volume *vol = NTFS_SB(sb);
444
445    ntfs_debug("Entering with remount options string: %s", opt);
446
447#ifndef NTFS_RW
448    /* For read-only compiled driver, enforce read-only flag. */
449    *flags |= MS_RDONLY;
450#else /* NTFS_RW */
451    /*
452     * For the read-write compiled driver, if we are remounting read-write,
453     * make sure there are no volume errors and that no unsupported volume
454     * flags are set. Also, empty the logfile journal as it would become
455     * stale as soon as something is written to the volume and mark the
456     * volume dirty so that chkdsk is run if the volume is not umounted
457     * cleanly. Finally, mark the quotas out of date so Windows rescans
458     * the volume on boot and updates them.
459     *
460     * When remounting read-only, mark the volume clean if no volume errors
461     * have occurred.
462     */
463    if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
464        static const char *es = ". Cannot remount read-write.";
465
466        /* Remounting read-write. */
467        if (NVolErrors(vol)) {
468            ntfs_error(sb, "Volume has errors and is read-only%s",
469                    es);
470            return -EROFS;
471        }
472        if (vol->vol_flags & VOLUME_IS_DIRTY) {
473            ntfs_error(sb, "Volume is dirty and read-only%s", es);
474            return -EROFS;
475        }
476        if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
477            ntfs_error(sb, "Volume has been modified by chkdsk "
478                    "and is read-only%s", es);
479            return -EROFS;
480        }
481        if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
482            ntfs_error(sb, "Volume has unsupported flags set "
483                    "(0x%x) and is read-only%s",
484                    (unsigned)le16_to_cpu(vol->vol_flags),
485                    es);
486            return -EROFS;
487        }
488        if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
489            ntfs_error(sb, "Failed to set dirty bit in volume "
490                    "information flags%s", es);
491            return -EROFS;
492        }
493#if 0
494        // TODO: Enable this code once we start modifying anything that
495        // is different between NTFS 1.2 and 3.x...
496        /* Set NT4 compatibility flag on newer NTFS version volumes. */
497        if ((vol->major_ver > 1)) {
498            if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
499                ntfs_error(sb, "Failed to set NT4 "
500                        "compatibility flag%s", es);
501                NVolSetErrors(vol);
502                return -EROFS;
503            }
504        }
505#endif
506        if (!ntfs_empty_logfile(vol->logfile_ino)) {
507            ntfs_error(sb, "Failed to empty journal $LogFile%s",
508                    es);
509            NVolSetErrors(vol);
510            return -EROFS;
511        }
512        if (!ntfs_mark_quotas_out_of_date(vol)) {
513            ntfs_error(sb, "Failed to mark quotas out of date%s",
514                    es);
515            NVolSetErrors(vol);
516            return -EROFS;
517        }
518        if (!ntfs_stamp_usnjrnl(vol)) {
519            ntfs_error(sb, "Failed to stamp transation log "
520                    "($UsnJrnl)%s", es);
521            NVolSetErrors(vol);
522            return -EROFS;
523        }
524    } else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
525        /* Remounting read-only. */
526        if (!NVolErrors(vol)) {
527            if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
528                ntfs_warning(sb, "Failed to clear dirty bit "
529                        "in volume information "
530                        "flags. Run chkdsk.");
531        }
532    }
533#endif /* NTFS_RW */
534
535    // TODO: Deal with *flags.
536
537    if (!parse_options(vol, opt))
538        return -EINVAL;
539
540    ntfs_debug("Done.");
541    return 0;
542}
543
544/**
545 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
546 * @sb: Super block of the device to which @b belongs.
547 * @b: Boot sector of device @sb to check.
548 * @silent: If 'true', all output will be silenced.
549 *
550 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
551 * sector. Returns 'true' if it is valid and 'false' if not.
552 *
553 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
554 * is 'true'.
555 */
556static bool is_boot_sector_ntfs(const struct super_block *sb,
557        const NTFS_BOOT_SECTOR *b, const bool silent)
558{
559    /*
560     * Check that checksum == sum of u32 values from b to the checksum
561     * field. If checksum is zero, no checking is done. We will work when
562     * the checksum test fails, since some utilities update the boot sector
563     * ignoring the checksum which leaves the checksum out-of-date. We
564     * report a warning if this is the case.
565     */
566    if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
567        le32 *u;
568        u32 i;
569
570        for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
571            i += le32_to_cpup(u);
572        if (le32_to_cpu(b->checksum) != i)
573            ntfs_warning(sb, "Invalid boot sector checksum.");
574    }
575    /* Check OEMidentifier is "NTFS " */
576    if (b->oem_id != magicNTFS)
577        goto not_ntfs;
578    /* Check bytes per sector value is between 256 and 4096. */
579    if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
580            le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
581        goto not_ntfs;
582    /* Check sectors per cluster value is valid. */
583    switch (b->bpb.sectors_per_cluster) {
584    case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
585        break;
586    default:
587        goto not_ntfs;
588    }
589    /* Check the cluster size is not above the maximum (64kiB). */
590    if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
591            b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
592        goto not_ntfs;
593    /* Check reserved/unused fields are really zero. */
594    if (le16_to_cpu(b->bpb.reserved_sectors) ||
595            le16_to_cpu(b->bpb.root_entries) ||
596            le16_to_cpu(b->bpb.sectors) ||
597            le16_to_cpu(b->bpb.sectors_per_fat) ||
598            le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
599        goto not_ntfs;
600    /* Check clusters per file mft record value is valid. */
601    if ((u8)b->clusters_per_mft_record < 0xe1 ||
602            (u8)b->clusters_per_mft_record > 0xf7)
603        switch (b->clusters_per_mft_record) {
604        case 1: case 2: case 4: case 8: case 16: case 32: case 64:
605            break;
606        default:
607            goto not_ntfs;
608        }
609    /* Check clusters per index block value is valid. */
610    if ((u8)b->clusters_per_index_record < 0xe1 ||
611            (u8)b->clusters_per_index_record > 0xf7)
612        switch (b->clusters_per_index_record) {
613        case 1: case 2: case 4: case 8: case 16: case 32: case 64:
614            break;
615        default:
616            goto not_ntfs;
617        }
618    /*
619     * Check for valid end of sector marker. We will work without it, but
620     * many BIOSes will refuse to boot from a bootsector if the magic is
621     * incorrect, so we emit a warning.
622     */
623    if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
624        ntfs_warning(sb, "Invalid end of sector marker.");
625    return true;
626not_ntfs:
627    return false;
628}
629
630/**
631 * read_ntfs_boot_sector - read the NTFS boot sector of a device
632 * @sb: super block of device to read the boot sector from
633 * @silent: if true, suppress all output
634 *
635 * Reads the boot sector from the device and validates it. If that fails, tries
636 * to read the backup boot sector, first from the end of the device a-la NT4 and
637 * later and then from the middle of the device a-la NT3.51 and before.
638 *
639 * If a valid boot sector is found but it is not the primary boot sector, we
640 * repair the primary boot sector silently (unless the device is read-only or
641 * the primary boot sector is not accessible).
642 *
643 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
644 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
645 * to their respective values.
646 *
647 * Return the unlocked buffer head containing the boot sector or NULL on error.
648 */
649static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
650        const int silent)
651{
652    const char *read_err_str = "Unable to read %s boot sector.";
653    struct buffer_head *bh_primary, *bh_backup;
654    sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
655
656    /* Try to read primary boot sector. */
657    if ((bh_primary = sb_bread(sb, 0))) {
658        if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
659                bh_primary->b_data, silent))
660            return bh_primary;
661        if (!silent)
662            ntfs_error(sb, "Primary boot sector is invalid.");
663    } else if (!silent)
664        ntfs_error(sb, read_err_str, "primary");
665    if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
666        if (bh_primary)
667            brelse(bh_primary);
668        if (!silent)
669            ntfs_error(sb, "Mount option errors=recover not used. "
670                    "Aborting without trying to recover.");
671        return NULL;
672    }
673    /* Try to read NT4+ backup boot sector. */
674    if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
675        if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
676                bh_backup->b_data, silent))
677            goto hotfix_primary_boot_sector;
678        brelse(bh_backup);
679    } else if (!silent)
680        ntfs_error(sb, read_err_str, "backup");
681    /* Try to read NT3.51- backup boot sector. */
682    if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
683        if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
684                bh_backup->b_data, silent))
685            goto hotfix_primary_boot_sector;
686        if (!silent)
687            ntfs_error(sb, "Could not find a valid backup boot "
688                    "sector.");
689        brelse(bh_backup);
690    } else if (!silent)
691        ntfs_error(sb, read_err_str, "backup");
692    /* We failed. Cleanup and return. */
693    if (bh_primary)
694        brelse(bh_primary);
695    return NULL;
696hotfix_primary_boot_sector:
697    if (bh_primary) {
698        /*
699         * If we managed to read sector zero and the volume is not
700         * read-only, copy the found, valid backup boot sector to the
701         * primary boot sector. Note we only copy the actual boot
702         * sector structure, not the actual whole device sector as that
703         * may be bigger and would potentially damage the $Boot system
704         * file (FIXME: Would be nice to know if the backup boot sector
705         * on a large sector device contains the whole boot loader or
706         * just the first 512 bytes).
707         */
708        if (!(sb->s_flags & MS_RDONLY)) {
709            ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
710                    "boot sector from backup copy.");
711            memcpy(bh_primary->b_data, bh_backup->b_data,
712                    NTFS_BLOCK_SIZE);
713            mark_buffer_dirty(bh_primary);
714            sync_dirty_buffer(bh_primary);
715            if (buffer_uptodate(bh_primary)) {
716                brelse(bh_backup);
717                return bh_primary;
718            }
719            ntfs_error(sb, "Hot-fix: Device write error while "
720                    "recovering primary boot sector.");
721        } else {
722            ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
723                    "sector failed: Read-only mount.");
724        }
725        brelse(bh_primary);
726    }
727    ntfs_warning(sb, "Using backup boot sector.");
728    return bh_backup;
729}
730
731/**
732 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
733 * @vol: volume structure to initialise with data from boot sector
734 * @b: boot sector to parse
735 *
736 * Parse the ntfs boot sector @b and store all imporant information therein in
737 * the ntfs super block @vol. Return 'true' on success and 'false' on error.
738 */
739static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
740{
741    unsigned int sectors_per_cluster_bits, nr_hidden_sects;
742    int clusters_per_mft_record, clusters_per_index_record;
743    s64 ll;
744
745    vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
746    vol->sector_size_bits = ffs(vol->sector_size) - 1;
747    ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
748            vol->sector_size);
749    ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
750            vol->sector_size_bits);
751    if (vol->sector_size < vol->sb->s_blocksize) {
752        ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
753                "device block size (%lu). This is not "
754                "supported. Sorry.", vol->sector_size,
755                vol->sb->s_blocksize);
756        return false;
757    }
758    ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
759    sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
760    ntfs_debug("sectors_per_cluster_bits = 0x%x",
761            sectors_per_cluster_bits);
762    nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
763    ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
764    vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
765    vol->cluster_size_mask = vol->cluster_size - 1;
766    vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
767    ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
768            vol->cluster_size);
769    ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
770    ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
771    if (vol->cluster_size < vol->sector_size) {
772        ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
773                "sector size (%i). This is not supported. "
774                "Sorry.", vol->cluster_size, vol->sector_size);
775        return false;
776    }
777    clusters_per_mft_record = b->clusters_per_mft_record;
778    ntfs_debug("clusters_per_mft_record = %i (0x%x)",
779            clusters_per_mft_record, clusters_per_mft_record);
780    if (clusters_per_mft_record > 0)
781        vol->mft_record_size = vol->cluster_size <<
782                (ffs(clusters_per_mft_record) - 1);
783    else
784        /*
785         * When mft_record_size < cluster_size, clusters_per_mft_record
786         * = -log2(mft_record_size) bytes. mft_record_size normaly is
787         * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
788         */
789        vol->mft_record_size = 1 << -clusters_per_mft_record;
790    vol->mft_record_size_mask = vol->mft_record_size - 1;
791    vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
792    ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
793            vol->mft_record_size);
794    ntfs_debug("vol->mft_record_size_mask = 0x%x",
795            vol->mft_record_size_mask);
796    ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
797            vol->mft_record_size_bits, vol->mft_record_size_bits);
798    /*
799     * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
800     * we store $MFT/$DATA, the table of mft records in the page cache.
801     */
802    if (vol->mft_record_size > PAGE_CACHE_SIZE) {
803        ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
804                "PAGE_CACHE_SIZE on your system (%lu). "
805                "This is not supported. Sorry.",
806                vol->mft_record_size, PAGE_CACHE_SIZE);
807        return false;
808    }
809    /* We cannot support mft record sizes below the sector size. */
810    if (vol->mft_record_size < vol->sector_size) {
811        ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
812                "sector size (%i). This is not supported. "
813                "Sorry.", vol->mft_record_size,
814                vol->sector_size);
815        return false;
816    }
817    clusters_per_index_record = b->clusters_per_index_record;
818    ntfs_debug("clusters_per_index_record = %i (0x%x)",
819            clusters_per_index_record, clusters_per_index_record);
820    if (clusters_per_index_record > 0)
821        vol->index_record_size = vol->cluster_size <<
822                (ffs(clusters_per_index_record) - 1);
823    else
824        /*
825         * When index_record_size < cluster_size,
826         * clusters_per_index_record = -log2(index_record_size) bytes.
827         * index_record_size normaly equals 4096 bytes, which is
828         * encoded as 0xF4 (-12 in decimal).
829         */
830        vol->index_record_size = 1 << -clusters_per_index_record;
831    vol->index_record_size_mask = vol->index_record_size - 1;
832    vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
833    ntfs_debug("vol->index_record_size = %i (0x%x)",
834            vol->index_record_size, vol->index_record_size);
835    ntfs_debug("vol->index_record_size_mask = 0x%x",
836            vol->index_record_size_mask);
837    ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
838            vol->index_record_size_bits,
839            vol->index_record_size_bits);
840    /* We cannot support index record sizes below the sector size. */
841    if (vol->index_record_size < vol->sector_size) {
842        ntfs_error(vol->sb, "Index record size (%i) is smaller than "
843                "the sector size (%i). This is not "
844                "supported. Sorry.", vol->index_record_size,
845                vol->sector_size);
846        return false;
847    }
848    /*
849     * Get the size of the volume in clusters and check for 64-bit-ness.
850     * Windows currently only uses 32 bits to save the clusters so we do
851     * the same as it is much faster on 32-bit CPUs.
852     */
853    ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
854    if ((u64)ll >= 1ULL << 32) {
855        ntfs_error(vol->sb, "Cannot handle 64-bit clusters. Sorry.");
856        return false;
857    }
858    vol->nr_clusters = ll;
859    ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
860    /*
861     * On an architecture where unsigned long is 32-bits, we restrict the
862     * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
863     * will hopefully optimize the whole check away.
864     */
865    if (sizeof(unsigned long) < 8) {
866        if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
867            ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
868                    "large for this architecture. "
869                    "Maximum supported is 2TiB. Sorry.",
870                    (unsigned long long)ll >> (40 -
871                    vol->cluster_size_bits));
872            return false;
873        }
874    }
875    ll = sle64_to_cpu(b->mft_lcn);
876    if (ll >= vol->nr_clusters) {
877        ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
878                "volume. Weird.", (unsigned long long)ll,
879                (unsigned long long)ll);
880        return false;
881    }
882    vol->mft_lcn = ll;
883    ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
884    ll = sle64_to_cpu(b->mftmirr_lcn);
885    if (ll >= vol->nr_clusters) {
886        ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
887                "of volume. Weird.", (unsigned long long)ll,
888                (unsigned long long)ll);
889        return false;
890    }
891    vol->mftmirr_lcn = ll;
892    ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
893#ifdef NTFS_RW
894    /*
895     * Work out the size of the mft mirror in number of mft records. If the
896     * cluster size is less than or equal to the size taken by four mft
897     * records, the mft mirror stores the first four mft records. If the
898     * cluster size is bigger than the size taken by four mft records, the
899     * mft mirror contains as many mft records as will fit into one
900     * cluster.
901     */
902    if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
903        vol->mftmirr_size = 4;
904    else
905        vol->mftmirr_size = vol->cluster_size >>
906                vol->mft_record_size_bits;
907    ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
908#endif /* NTFS_RW */
909    vol->serial_no = le64_to_cpu(b->volume_serial_number);
910    ntfs_debug("vol->serial_no = 0x%llx",
911            (unsigned long long)vol->serial_no);
912    return true;
913}
914
915/**
916 * ntfs_setup_allocators - initialize the cluster and mft allocators
917 * @vol: volume structure for which to setup the allocators
918 *
919 * Setup the cluster (lcn) and mft allocators to the starting values.
920 */
921static void ntfs_setup_allocators(ntfs_volume *vol)
922{
923#ifdef NTFS_RW
924    LCN mft_zone_size, mft_lcn;
925#endif /* NTFS_RW */
926
927    ntfs_debug("vol->mft_zone_multiplier = 0x%x",
928            vol->mft_zone_multiplier);
929#ifdef NTFS_RW
930    /* Determine the size of the MFT zone. */
931    mft_zone_size = vol->nr_clusters;
932    switch (vol->mft_zone_multiplier) { /* % of volume size in clusters */
933    case 4:
934        mft_zone_size >>= 1; /* 50% */
935        break;
936    case 3:
937        mft_zone_size = (mft_zone_size +
938                (mft_zone_size >> 1)) >> 2; /* 37.5% */
939        break;
940    case 2:
941        mft_zone_size >>= 2; /* 25% */
942        break;
943    /* case 1: */
944    default:
945        mft_zone_size >>= 3; /* 12.5% */
946        break;
947    }
948    /* Setup the mft zone. */
949    vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
950    ntfs_debug("vol->mft_zone_pos = 0x%llx",
951            (unsigned long long)vol->mft_zone_pos);
952    /*
953     * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
954     * source) and if the actual mft_lcn is in the expected place or even
955     * further to the front of the volume, extend the mft_zone to cover the
956     * beginning of the volume as well. This is in order to protect the
957     * area reserved for the mft bitmap as well within the mft_zone itself.
958     * On non-standard volumes we do not protect it as the overhead would
959     * be higher than the speed increase we would get by doing it.
960     */
961    mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
962    if (mft_lcn * vol->cluster_size < 16 * 1024)
963        mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
964                vol->cluster_size;
965    if (vol->mft_zone_start <= mft_lcn)
966        vol->mft_zone_start = 0;
967    ntfs_debug("vol->mft_zone_start = 0x%llx",
968            (unsigned long long)vol->mft_zone_start);
969    /*
970     * Need to cap the mft zone on non-standard volumes so that it does
971     * not point outside the boundaries of the volume. We do this by
972     * halving the zone size until we are inside the volume.
973     */
974    vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
975    while (vol->mft_zone_end >= vol->nr_clusters) {
976        mft_zone_size >>= 1;
977        vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
978    }
979    ntfs_debug("vol->mft_zone_end = 0x%llx",
980            (unsigned long long)vol->mft_zone_end);
981    /*
982     * Set the current position within each data zone to the start of the
983     * respective zone.
984     */
985    vol->data1_zone_pos = vol->mft_zone_end;
986    ntfs_debug("vol->data1_zone_pos = 0x%llx",
987            (unsigned long long)vol->data1_zone_pos);
988    vol->data2_zone_pos = 0;
989    ntfs_debug("vol->data2_zone_pos = 0x%llx",
990            (unsigned long long)vol->data2_zone_pos);
991
992    /* Set the mft data allocation position to mft record 24. */
993    vol->mft_data_pos = 24;
994    ntfs_debug("vol->mft_data_pos = 0x%llx",
995            (unsigned long long)vol->mft_data_pos);
996#endif /* NTFS_RW */
997}
998
999#ifdef NTFS_RW
1000
1001/**
1002 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
1003 * @vol: ntfs super block describing device whose mft mirror to load
1004 *
1005 * Return 'true' on success or 'false' on error.
1006 */
1007static bool load_and_init_mft_mirror(ntfs_volume *vol)
1008{
1009    struct inode *tmp_ino;
1010    ntfs_inode *tmp_ni;
1011
1012    ntfs_debug("Entering.");
1013    /* Get mft mirror inode. */
1014    tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
1015    if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1016        if (!IS_ERR(tmp_ino))
1017            iput(tmp_ino);
1018        /* Caller will display error message. */
1019        return false;
1020    }
1021    /*
1022     * Re-initialize some specifics about $MFTMirr's inode as
1023     * ntfs_read_inode() will have set up the default ones.
1024     */
1025    /* Set uid and gid to root. */
1026    tmp_ino->i_uid = tmp_ino->i_gid = 0;
1027    /* Regular file. No access for anyone. */
1028    tmp_ino->i_mode = S_IFREG;
1029    /* No VFS initiated operations allowed for $MFTMirr. */
1030    tmp_ino->i_op = &ntfs_empty_inode_ops;
1031    tmp_ino->i_fop = &ntfs_empty_file_ops;
1032    /* Put in our special address space operations. */
1033    tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1034    tmp_ni = NTFS_I(tmp_ino);
1035    /* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1036    NInoSetMstProtected(tmp_ni);
1037    NInoSetSparseDisabled(tmp_ni);
1038    /*
1039     * Set up our little cheat allowing us to reuse the async read io
1040     * completion handler for directories.
1041     */
1042    tmp_ni->itype.index.block_size = vol->mft_record_size;
1043    tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1044    vol->mftmirr_ino = tmp_ino;
1045    ntfs_debug("Done.");
1046    return true;
1047}
1048
1049/**
1050 * check_mft_mirror - compare contents of the mft mirror with the mft
1051 * @vol: ntfs super block describing device whose mft mirror to check
1052 *
1053 * Return 'true' on success or 'false' on error.
1054 *
1055 * Note, this function also results in the mft mirror runlist being completely
1056 * mapped into memory. The mft mirror write code requires this and will BUG()
1057 * should it find an unmapped runlist element.
1058 */
1059static bool check_mft_mirror(ntfs_volume *vol)
1060{
1061    struct super_block *sb = vol->sb;
1062    ntfs_inode *mirr_ni;
1063    struct page *mft_page, *mirr_page;
1064    u8 *kmft, *kmirr;
1065    runlist_element *rl, rl2[2];
1066    pgoff_t index;
1067    int mrecs_per_page, i;
1068
1069    ntfs_debug("Entering.");
1070    /* Compare contents of $MFT and $MFTMirr. */
1071    mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1072    BUG_ON(!mrecs_per_page);
1073    BUG_ON(!vol->mftmirr_size);
1074    mft_page = mirr_page = NULL;
1075    kmft = kmirr = NULL;
1076    index = i = 0;
1077    do {
1078        u32 bytes;
1079
1080        /* Switch pages if necessary. */
1081        if (!(i % mrecs_per_page)) {
1082            if (index) {
1083                ntfs_unmap_page(mft_page);
1084                ntfs_unmap_page(mirr_page);
1085            }
1086            /* Get the $MFT page. */
1087            mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1088                    index);
1089            if (IS_ERR(mft_page)) {
1090                ntfs_error(sb, "Failed to read $MFT.");
1091                return false;
1092            }
1093            kmft = page_address(mft_page);
1094            /* Get the $MFTMirr page. */
1095            mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1096                    index);
1097            if (IS_ERR(mirr_page)) {
1098                ntfs_error(sb, "Failed to read $MFTMirr.");
1099                goto mft_unmap_out;
1100            }
1101            kmirr = page_address(mirr_page);
1102            ++index;
1103        }
1104        /* Do not check the record if it is not in use. */
1105        if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1106            /* Make sure the record is ok. */
1107            if (ntfs_is_baad_recordp((le32*)kmft)) {
1108                ntfs_error(sb, "Incomplete multi sector "
1109                        "transfer detected in mft "
1110                        "record %i.", i);
1111mm_unmap_out:
1112                ntfs_unmap_page(mirr_page);
1113mft_unmap_out:
1114                ntfs_unmap_page(mft_page);
1115                return false;
1116            }
1117        }
1118        /* Do not check the mirror record if it is not in use. */
1119        if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1120            if (ntfs_is_baad_recordp((le32*)kmirr)) {
1121                ntfs_error(sb, "Incomplete multi sector "
1122                        "transfer detected in mft "
1123                        "mirror record %i.", i);
1124                goto mm_unmap_out;
1125            }
1126        }
1127        /* Get the amount of data in the current record. */
1128        bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1129        if (bytes < sizeof(MFT_RECORD_OLD) ||
1130                bytes > vol->mft_record_size ||
1131                ntfs_is_baad_recordp((le32*)kmft)) {
1132            bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1133            if (bytes < sizeof(MFT_RECORD_OLD) ||
1134                    bytes > vol->mft_record_size ||
1135                    ntfs_is_baad_recordp((le32*)kmirr))
1136                bytes = vol->mft_record_size;
1137        }
1138        /* Compare the two records. */
1139        if (memcmp(kmft, kmirr, bytes)) {
1140            ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1141                    "match. Run ntfsfix or chkdsk.", i);
1142            goto mm_unmap_out;
1143        }
1144        kmft += vol->mft_record_size;
1145        kmirr += vol->mft_record_size;
1146    } while (++i < vol->mftmirr_size);
1147    /* Release the last pages. */
1148    ntfs_unmap_page(mft_page);
1149    ntfs_unmap_page(mirr_page);
1150
1151    /* Construct the mft mirror runlist by hand. */
1152    rl2[0].vcn = 0;
1153    rl2[0].lcn = vol->mftmirr_lcn;
1154    rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1155            vol->cluster_size - 1) / vol->cluster_size;
1156    rl2[1].vcn = rl2[0].length;
1157    rl2[1].lcn = LCN_ENOENT;
1158    rl2[1].length = 0;
1159    /*
1160     * Because we have just read all of the mft mirror, we know we have
1161     * mapped the full runlist for it.
1162     */
1163    mirr_ni = NTFS_I(vol->mftmirr_ino);
1164    down_read(&mirr_ni->runlist.lock);
1165    rl = mirr_ni->runlist.rl;
1166    /* Compare the two runlists. They must be identical. */
1167    i = 0;
1168    do {
1169        if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1170                rl2[i].length != rl[i].length) {
1171            ntfs_error(sb, "$MFTMirr location mismatch. "
1172                    "Run chkdsk.");
1173            up_read(&mirr_ni->runlist.lock);
1174            return false;
1175        }
1176    } while (rl2[i++].length);
1177    up_read(&mirr_ni->runlist.lock);
1178    ntfs_debug("Done.");
1179    return true;
1180}
1181
1182/**
1183 * load_and_check_logfile - load and check the logfile inode for a volume
1184 * @vol: ntfs super block describing device whose logfile to load
1185 *
1186 * Return 'true' on success or 'false' on error.
1187 */
1188static bool load_and_check_logfile(ntfs_volume *vol,
1189        RESTART_PAGE_HEADER **rp)
1190{
1191    struct inode *tmp_ino;
1192
1193    ntfs_debug("Entering.");
1194    tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1195    if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1196        if (!IS_ERR(tmp_ino))
1197            iput(tmp_ino);
1198        /* Caller will display error message. */
1199        return false;
1200    }
1201    if (!ntfs_check_logfile(tmp_ino, rp)) {
1202        iput(tmp_ino);
1203        /* ntfs_check_logfile() will have displayed error output. */
1204        return false;
1205    }
1206    NInoSetSparseDisabled(NTFS_I(tmp_ino));
1207    vol->logfile_ino = tmp_ino;
1208    ntfs_debug("Done.");
1209    return true;
1210}
1211
1212#define NTFS_HIBERFIL_HEADER_SIZE 4096
1213
1214/**
1215 * check_windows_hibernation_status - check if Windows is suspended on a volume
1216 * @vol: ntfs super block of device to check
1217 *
1218 * Check if Windows is hibernated on the ntfs volume @vol. This is done by
1219 * looking for the file hiberfil.sys in the root directory of the volume. If
1220 * the file is not present Windows is definitely not suspended.
1221 *
1222 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1223 * definitely suspended (this volume is not the system volume). Caveat: on a
1224 * system with many volumes it is possible that the < 4kiB check is bogus but
1225 * for now this should do fine.
1226 *
1227 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1228 * hiberfil header (which is the first 4kiB). If this begins with "hibr",
1229 * Windows is definitely suspended. If it is completely full of zeroes,
1230 * Windows is definitely not hibernated. Any other case is treated as if
1231 * Windows is suspended. This caters for the above mentioned caveat of a
1232 * system with many volumes where no "hibr" magic would be present and there is
1233 * no zero header.
1234 *
1235 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1236 * hibernated on the volume, and -errno on error.
1237 */
1238static int check_windows_hibernation_status(ntfs_volume *vol)
1239{
1240    MFT_REF mref;
1241    struct inode *vi;
1242    ntfs_inode *ni;
1243    struct page *page;
1244    u32 *kaddr, *kend;
1245    ntfs_name *name = NULL;
1246    int ret = 1;
1247    static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
1248            cpu_to_le16('i'), cpu_to_le16('b'),
1249            cpu_to_le16('e'), cpu_to_le16('r'),
1250            cpu_to_le16('f'), cpu_to_le16('i'),
1251            cpu_to_le16('l'), cpu_to_le16('.'),
1252            cpu_to_le16('s'), cpu_to_le16('y'),
1253            cpu_to_le16('s'), 0 };
1254
1255    ntfs_debug("Entering.");
1256    /*
1257     * Find the inode number for the hibernation file by looking up the
1258     * filename hiberfil.sys in the root directory.
1259     */
1260    mutex_lock(&vol->root_ino->i_mutex);
1261    mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1262            &name);
1263    mutex_unlock(&vol->root_ino->i_mutex);
1264    if (IS_ERR_MREF(mref)) {
1265        ret = MREF_ERR(mref);
1266        /* If the file does not exist, Windows is not hibernated. */
1267        if (ret == -ENOENT) {
1268            ntfs_debug("hiberfil.sys not present. Windows is not "
1269                    "hibernated on the volume.");
1270            return 0;
1271        }
1272        /* A real error occurred. */
1273        ntfs_error(vol->sb, "Failed to find inode number for "
1274                "hiberfil.sys.");
1275        return ret;
1276    }
1277    /* We do not care for the type of match that was found. */
1278    kfree(name);
1279    /* Get the inode. */
1280    vi = ntfs_iget(vol->sb, MREF(mref));
1281    if (IS_ERR(vi) || is_bad_inode(vi)) {
1282        if (!IS_ERR(vi))
1283            iput(vi);
1284        ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1285        return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1286    }
1287    if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1288        ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx). "
1289                "Windows is hibernated on the volume. This "
1290                "is not the system volume.", i_size_read(vi));
1291        goto iput_out;
1292    }
1293    ni = NTFS_I(vi);
1294    page = ntfs_map_page(vi->i_mapping, 0);
1295    if (IS_ERR(page)) {
1296        ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1297        ret = PTR_ERR(page);
1298        goto iput_out;
1299    }
1300    kaddr = (u32*)page_address(page);
1301    if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
1302        ntfs_debug("Magic \"hibr\" found in hiberfil.sys. Windows is "
1303                "hibernated on the volume. This is the "
1304                "system volume.");
1305        goto unm_iput_out;
1306    }
1307    kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1308    do {
1309        if (unlikely(*kaddr)) {
1310            ntfs_debug("hiberfil.sys is larger than 4kiB "
1311                    "(0x%llx), does not contain the "
1312                    "\"hibr\" magic, and does not have a "
1313                    "zero header. Windows is hibernated "
1314                    "on the volume. This is not the "
1315                    "system volume.", i_size_read(vi));
1316            goto unm_iput_out;
1317        }
1318    } while (++kaddr < kend);
1319    ntfs_debug("hiberfil.sys contains a zero header. Windows is not "
1320            "hibernated on the volume. This is the system "
1321            "volume.");
1322    ret = 0;
1323unm_iput_out:
1324    ntfs_unmap_page(page);
1325iput_out:
1326    iput(vi);
1327    return ret;
1328}
1329
1330/**
1331 * load_and_init_quota - load and setup the quota file for a volume if present
1332 * @vol: ntfs super block describing device whose quota file to load
1333 *
1334 * Return 'true' on success or 'false' on error. If $Quota is not present, we
1335 * leave vol->quota_ino as NULL and return success.
1336 */
1337static bool load_and_init_quota(ntfs_volume *vol)
1338{
1339    MFT_REF mref;
1340    struct inode *tmp_ino;
1341    ntfs_name *name = NULL;
1342    static const ntfschar Quota[7] = { cpu_to_le16('$'),
1343            cpu_to_le16('Q'), cpu_to_le16('u'),
1344            cpu_to_le16('o'), cpu_to_le16('t'),
1345            cpu_to_le16('a'), 0 };
1346    static ntfschar Q[3] = { cpu_to_le16('$'),
1347            cpu_to_le16('Q'), 0 };
1348
1349    ntfs_debug("Entering.");
1350    /*
1351     * Find the inode number for the quota file by looking up the filename
1352     * $Quota in the extended system files directory $Extend.
1353     */
1354    mutex_lock(&vol->extend_ino->i_mutex);
1355    mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1356            &name);
1357    mutex_unlock(&vol->extend_ino->i_mutex);
1358    if (IS_ERR_MREF(mref)) {
1359        /*
1360         * If the file does not exist, quotas are disabled and have
1361         * never been enabled on this volume, just return success.
1362         */
1363        if (MREF_ERR(mref) == -ENOENT) {
1364            ntfs_debug("$Quota not present. Volume does not have "
1365                    "quotas enabled.");
1366            /*
1367             * No need to try to set quotas out of date if they are
1368             * not enabled.
1369             */
1370            NVolSetQuotaOutOfDate(vol);
1371            return true;
1372        }
1373        /* A real error occurred. */
1374        ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1375        return false;
1376    }
1377    /* We do not care for the type of match that was found. */
1378    kfree(name);
1379    /* Get the inode. */
1380    tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1381    if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1382        if (!IS_ERR(tmp_ino))
1383            iput(tmp_ino);
1384        ntfs_error(vol->sb, "Failed to load $Quota.");
1385        return false;
1386    }
1387    vol->quota_ino = tmp_ino;
1388    /* Get the $Q index allocation attribute. */
1389    tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1390    if (IS_ERR(tmp_ino)) {
1391        ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1392        return false;
1393    }
1394    vol->quota_q_ino = tmp_ino;
1395    ntfs_debug("Done.");
1396    return true;
1397}
1398
1399/**
1400 * load_and_init_usnjrnl - load and setup the transaction log if present
1401 * @vol: ntfs super block describing device whose usnjrnl file to load
1402 *
1403 * Return 'true' on success or 'false' on error.
1404 *
1405 * If $UsnJrnl is not present or in the process of being disabled, we set
1406 * NVolUsnJrnlStamped() and return success.
1407 *
1408 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1409 * i.e. transaction logging has only just been enabled or the journal has been
1410 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1411 * and return success.
1412 */
1413static bool load_and_init_usnjrnl(ntfs_volume *vol)
1414{
1415    MFT_REF mref;
1416    struct inode *tmp_ino;
1417    ntfs_inode *tmp_ni;
1418    struct page *page;
1419    ntfs_name *name = NULL;
1420    USN_HEADER *uh;
1421    static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
1422            cpu_to_le16('U'), cpu_to_le16('s'),
1423            cpu_to_le16('n'), cpu_to_le16('J'),
1424            cpu_to_le16('r'), cpu_to_le16('n'),
1425            cpu_to_le16('l'), 0 };
1426    static ntfschar Max[5] = { cpu_to_le16('$'),
1427            cpu_to_le16('M'), cpu_to_le16('a'),
1428            cpu_to_le16('x'), 0 };
1429    static ntfschar J[3] = { cpu_to_le16('$'),
1430            cpu_to_le16('J'), 0 };
1431
1432    ntfs_debug("Entering.");
1433    /*
1434     * Find the inode number for the transaction log file by looking up the
1435     * filename $UsnJrnl in the extended system files directory $Extend.
1436     */
1437    mutex_lock(&vol->extend_ino->i_mutex);
1438    mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1439            &name);
1440    mutex_unlock(&vol->extend_ino->i_mutex);
1441    if (IS_ERR_MREF(mref)) {
1442        /*
1443         * If the file does not exist, transaction logging is disabled,
1444         * just return success.
1445         */
1446        if (MREF_ERR(mref) == -ENOENT) {
1447            ntfs_debug("$UsnJrnl not present. Volume does not "
1448                    "have transaction logging enabled.");
1449not_enabled:
1450            /*
1451             * No need to try to stamp the transaction log if
1452             * transaction logging is not enabled.
1453             */
1454            NVolSetUsnJrnlStamped(vol);
1455            return true;
1456        }
1457        /* A real error occurred. */
1458        ntfs_error(vol->sb, "Failed to find inode number for "
1459                "$UsnJrnl.");
1460        return false;
1461    }
1462    /* We do not care for the type of match that was found. */
1463    kfree(name);
1464    /* Get the inode. */
1465    tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1466    if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1467        if (!IS_ERR(tmp_ino))
1468            iput(tmp_ino);
1469        ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1470        return false;
1471    }
1472    vol->usnjrnl_ino = tmp_ino;
1473    /*
1474     * If the transaction log is in the process of being deleted, we can
1475     * ignore it.
1476     */
1477    if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1478        ntfs_debug("$UsnJrnl in the process of being disabled. "
1479                "Volume does not have transaction logging "
1480                "enabled.");
1481        goto not_enabled;
1482    }
1483    /* Get the $DATA/$Max attribute. */
1484    tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1485    if (IS_ERR(tmp_ino)) {
1486        ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1487                "attribute.");
1488        return false;
1489    }
1490    vol->usnjrnl_max_ino = tmp_ino;
1491    if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1492        ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1493                "attribute (size is 0x%llx but should be at "
1494                "least 0x%zx bytes).", i_size_read(tmp_ino),
1495                sizeof(USN_HEADER));
1496        return false;
1497    }
1498    /* Get the $DATA/$J attribute. */
1499    tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1500    if (IS_ERR(tmp_ino)) {
1501        ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1502                "attribute.");
1503        return false;
1504    }
1505    vol->usnjrnl_j_ino = tmp_ino;
1506    /* Verify $J is non-resident and sparse. */
1507    tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1508    if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1509        ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1510                "and/or not sparse.");
1511        return false;
1512    }
1513    /* Read the USN_HEADER from $DATA/$Max. */
1514    page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1515    if (IS_ERR(page)) {
1516        ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1517                "attribute.");
1518        return false;
1519    }
1520    uh = (USN_HEADER*)page_address(page);
1521    /* Sanity check the $Max. */
1522    if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1523            sle64_to_cpu(uh->maximum_size))) {
1524        ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1525                "maximum size (0x%llx). $UsnJrnl is corrupt.",
1526                (long long)sle64_to_cpu(uh->allocation_delta),
1527                (long long)sle64_to_cpu(uh->maximum_size));
1528        ntfs_unmap_page(page);
1529        return false;
1530    }
1531    /*
1532     * If the transaction log has been stamped and nothing has been written
1533     * to it since, we do not need to stamp it.
1534     */
1535    if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1536            i_size_read(vol->usnjrnl_j_ino))) {
1537        if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1538                i_size_read(vol->usnjrnl_j_ino))) {
1539            ntfs_unmap_page(page);
1540            ntfs_debug("$UsnJrnl is enabled but nothing has been "
1541                    "logged since it was last stamped. "
1542                    "Treating this as if the volume does "
1543                    "not have transaction logging "
1544                    "enabled.");
1545            goto not_enabled;
1546        }
1547        ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1548                "which is out of bounds (0x%llx). $UsnJrnl "
1549                "is corrupt.",
1550                (long long)sle64_to_cpu(uh->lowest_valid_usn),
1551                i_size_read(vol->usnjrnl_j_ino));
1552        ntfs_unmap_page(page);
1553        return false;
1554    }
1555    ntfs_unmap_page(page);
1556    ntfs_debug("Done.");
1557    return true;
1558}
1559
1560/**
1561 * load_and_init_attrdef - load the attribute definitions table for a volume
1562 * @vol: ntfs super block describing device whose attrdef to load
1563 *
1564 * Return 'true' on success or 'false' on error.
1565 */
1566static bool load_and_init_attrdef(ntfs_volume *vol)
1567{
1568    loff_t i_size;
1569    struct super_block *sb = vol->sb;
1570    struct inode *ino;
1571    struct page *page;
1572    pgoff_t index, max_index;
1573    unsigned int size;
1574
1575    ntfs_debug("Entering.");
1576    /* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1577    ino = ntfs_iget(sb, FILE_AttrDef);
1578    if (IS_ERR(ino) || is_bad_inode(ino)) {
1579        if (!IS_ERR(ino))
1580            iput(ino);
1581        goto failed;
1582    }
1583    NInoSetSparseDisabled(NTFS_I(ino));
1584    /* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1585    i_size = i_size_read(ino);
1586    if (i_size <= 0 || i_size > 0x7fffffff)
1587        goto iput_failed;
1588    vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1589    if (!vol->attrdef)
1590        goto iput_failed;
1591    index = 0;
1592    max_index = i_size >> PAGE_CACHE_SHIFT;
1593    size = PAGE_CACHE_SIZE;
1594    while (index < max_index) {
1595        /* Read the attrdef table and copy it into the linear buffer. */
1596read_partial_attrdef_page:
1597        page = ntfs_map_page(ino->i_mapping, index);
1598        if (IS_ERR(page))
1599            goto free_iput_failed;
1600        memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1601                page_address(page), size);
1602        ntfs_unmap_page(page);
1603    };
1604    if (size == PAGE_CACHE_SIZE) {
1605        size = i_size & ~PAGE_CACHE_MASK;
1606        if (size)
1607            goto read_partial_attrdef_page;
1608    }
1609    vol->attrdef_size = i_size;
1610    ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1611    iput(ino);
1612    return true;
1613free_iput_failed:
1614    ntfs_free(vol->attrdef);
1615    vol->attrdef = NULL;
1616iput_failed:
1617    iput(ino);
1618failed:
1619    ntfs_error(sb, "Failed to initialize attribute definition table.");
1620    return false;
1621}
1622
1623#endif /* NTFS_RW */
1624
1625/**
1626 * load_and_init_upcase - load the upcase table for an ntfs volume
1627 * @vol: ntfs super block describing device whose upcase to load
1628 *
1629 * Return 'true' on success or 'false' on error.
1630 */
1631static bool load_and_init_upcase(ntfs_volume *vol)
1632{
1633    loff_t i_size;
1634    struct super_block *sb = vol->sb;
1635    struct inode *ino;
1636    struct page *page;
1637    pgoff_t index, max_index;
1638    unsigned int size;
1639    int i, max;
1640
1641    ntfs_debug("Entering.");
1642    /* Read upcase table and setup vol->upcase and vol->upcase_len. */
1643    ino = ntfs_iget(sb, FILE_UpCase);
1644    if (IS_ERR(ino) || is_bad_inode(ino)) {
1645        if (!IS_ERR(ino))
1646            iput(ino);
1647        goto upcase_failed;
1648    }
1649    /*
1650     * The upcase size must not be above 64k Unicode characters, must not
1651     * be zero and must be a multiple of sizeof(ntfschar).
1652     */
1653    i_size = i_size_read(ino);
1654    if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1655            i_size > 64ULL * 1024 * sizeof(ntfschar))
1656        goto iput_upcase_failed;
1657    vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1658    if (!vol->upcase)
1659        goto iput_upcase_failed;
1660    index = 0;
1661    max_index = i_size >> PAGE_CACHE_SHIFT;
1662    size = PAGE_CACHE_SIZE;
1663    while (index < max_index) {
1664        /* Read the upcase table and copy it into the linear buffer. */
1665read_partial_upcase_page:
1666        page = ntfs_map_page(ino->i_mapping, index);
1667        if (IS_ERR(page))
1668            goto iput_upcase_failed;
1669        memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1670                page_address(page), size);
1671        ntfs_unmap_page(page);
1672    };
1673    if (size == PAGE_CACHE_SIZE) {
1674        size = i_size & ~PAGE_CACHE_MASK;
1675        if (size)
1676            goto read_partial_upcase_page;
1677    }
1678    vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1679    ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1680            i_size, 64 * 1024 * sizeof(ntfschar));
1681    iput(ino);
1682    mutex_lock(&ntfs_lock);
1683    if (!default_upcase) {
1684        ntfs_debug("Using volume specified $UpCase since default is "
1685                "not present.");
1686        mutex_unlock(&ntfs_lock);
1687        return true;
1688    }
1689    max = default_upcase_len;
1690    if (max > vol->upcase_len)
1691        max = vol->upcase_len;
1692    for (i = 0; i < max; i++)
1693        if (vol->upcase[i] != default_upcase[i])
1694            break;
1695    if (i == max) {
1696        ntfs_free(vol->upcase);
1697        vol->upcase = default_upcase;
1698        vol->upcase_len = max;
1699        ntfs_nr_upcase_users++;
1700        mutex_unlock(&ntfs_lock);
1701        ntfs_debug("Volume specified $UpCase matches default. Using "
1702                "default.");
1703        return true;
1704    }
1705    mutex_unlock(&ntfs_lock);
1706    ntfs_debug("Using volume specified $UpCase since it does not match "
1707            "the default.");
1708    return true;
1709iput_upcase_failed:
1710    iput(ino);
1711    ntfs_free(vol->upcase);
1712    vol->upcase = NULL;
1713upcase_failed:
1714    mutex_lock(&ntfs_lock);
1715    if (default_upcase) {
1716        vol->upcase = default_upcase;
1717        vol->upcase_len = default_upcase_len;
1718        ntfs_nr_upcase_users++;
1719        mutex_unlock(&ntfs_lock);
1720        ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1721                "default.");
1722        return true;
1723    }
1724    mutex_unlock(&ntfs_lock);
1725    ntfs_error(sb, "Failed to initialize upcase table.");
1726    return false;
1727}
1728
1729/*
1730 * The lcn and mft bitmap inodes are NTFS-internal inodes with
1731 * their own special locking rules:
1732 */
1733static struct lock_class_key
1734    lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1735    mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1736
1737/**
1738 * load_system_files - open the system files using normal functions
1739 * @vol: ntfs super block describing device whose system files to load
1740 *
1741 * Open the system files with normal access functions and complete setting up
1742 * the ntfs super block @vol.
1743 *
1744 * Return 'true' on success or 'false' on error.
1745 */
1746static bool load_system_files(ntfs_volume *vol)
1747{
1748    struct super_block *sb = vol->sb;
1749    MFT_RECORD *m;
1750    VOLUME_INFORMATION *vi;
1751    ntfs_attr_search_ctx *ctx;
1752#ifdef NTFS_RW
1753    RESTART_PAGE_HEADER *rp;
1754    int err;
1755#endif /* NTFS_RW */
1756
1757    ntfs_debug("Entering.");
1758#ifdef NTFS_RW
1759    /* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1760    if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1761        static const char *es1 = "Failed to load $MFTMirr";
1762        static const char *es2 = "$MFTMirr does not match $MFT";
1763        static const char *es3 = ". Run ntfsfix and/or chkdsk.";
1764
1765        /* If a read-write mount, convert it to a read-only mount. */
1766        if (!(sb->s_flags & MS_RDONLY)) {
1767            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1768                    ON_ERRORS_CONTINUE))) {
1769                ntfs_error(sb, "%s and neither on_errors="
1770                        "continue nor on_errors="
1771                        "remount-ro was specified%s",
1772                        !vol->mftmirr_ino ? es1 : es2,
1773                        es3);
1774                goto iput_mirr_err_out;
1775            }
1776            sb->s_flags |= MS_RDONLY;
1777            ntfs_error(sb, "%s. Mounting read-only%s",
1778                    !vol->mftmirr_ino ? es1 : es2, es3);
1779        } else
1780            ntfs_warning(sb, "%s. Will not be able to remount "
1781                    "read-write%s",
1782                    !vol->mftmirr_ino ? es1 : es2, es3);
1783        /* This will prevent a read-write remount. */
1784        NVolSetErrors(vol);
1785    }
1786#endif /* NTFS_RW */
1787    /* Get mft bitmap attribute inode. */
1788    vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1789    if (IS_ERR(vol->mftbmp_ino)) {
1790        ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1791        goto iput_mirr_err_out;
1792    }
1793    lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1794               &mftbmp_runlist_lock_key);
1795    lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1796               &mftbmp_mrec_lock_key);
1797    /* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1798    if (!load_and_init_upcase(vol))
1799        goto iput_mftbmp_err_out;
1800#ifdef NTFS_RW
1801    /*
1802     * Read attribute definitions table and setup @vol->attrdef and
1803     * @vol->attrdef_size.
1804     */
1805    if (!load_and_init_attrdef(vol))
1806        goto iput_upcase_err_out;
1807#endif /* NTFS_RW */
1808    /*
1809     * Get the cluster allocation bitmap inode and verify the size, no
1810     * need for any locking at this stage as we are already running
1811     * exclusively as we are mount in progress task.
1812     */
1813    vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1814    if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1815        if (!IS_ERR(vol->lcnbmp_ino))
1816            iput(vol->lcnbmp_ino);
1817        goto bitmap_failed;
1818    }
1819    lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1820               &lcnbmp_runlist_lock_key);
1821    lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1822               &lcnbmp_mrec_lock_key);
1823
1824    NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1825    if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1826        iput(vol->lcnbmp_ino);
1827bitmap_failed:
1828        ntfs_error(sb, "Failed to load $Bitmap.");
1829        goto iput_attrdef_err_out;
1830    }
1831    /*
1832     * Get the volume inode and setup our cache of the volume flags and
1833     * version.
1834     */
1835    vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1836    if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1837        if (!IS_ERR(vol->vol_ino))
1838            iput(vol->vol_ino);
1839volume_failed:
1840        ntfs_error(sb, "Failed to load $Volume.");
1841        goto iput_lcnbmp_err_out;
1842    }
1843    m = map_mft_record(NTFS_I(vol->vol_ino));
1844    if (IS_ERR(m)) {
1845iput_volume_failed:
1846        iput(vol->vol_ino);
1847        goto volume_failed;
1848    }
1849    if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1850        ntfs_error(sb, "Failed to get attribute search context.");
1851        goto get_ctx_vol_failed;
1852    }
1853    if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1854            ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1855err_put_vol:
1856        ntfs_attr_put_search_ctx(ctx);
1857get_ctx_vol_failed:
1858        unmap_mft_record(NTFS_I(vol->vol_ino));
1859        goto iput_volume_failed;
1860    }
1861    vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1862            le16_to_cpu(ctx->attr->data.resident.value_offset));
1863    /* Some bounds checks. */
1864    if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1865            le32_to_cpu(ctx->attr->data.resident.value_length) >
1866            (u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1867        goto err_put_vol;
1868    /* Copy the volume flags and version to the ntfs_volume structure. */
1869    vol->vol_flags = vi->flags;
1870    vol->major_ver = vi->major_ver;
1871    vol->minor_ver = vi->minor_ver;
1872    ntfs_attr_put_search_ctx(ctx);
1873    unmap_mft_record(NTFS_I(vol->vol_ino));
1874    printk(KERN_INFO "NTFS volume version %i.%i.\n", vol->major_ver,
1875            vol->minor_ver);
1876    if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1877        ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1878                "volume version %i.%i (need at least version "
1879                "3.0).", vol->major_ver, vol->minor_ver);
1880        NVolClearSparseEnabled(vol);
1881    }
1882#ifdef NTFS_RW
1883    /* Make sure that no unsupported volume flags are set. */
1884    if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1885        static const char *es1a = "Volume is dirty";
1886        static const char *es1b = "Volume has been modified by chkdsk";
1887        static const char *es1c = "Volume has unsupported flags set";
1888        static const char *es2a = ". Run chkdsk and mount in Windows.";
1889        static const char *es2b = ". Mount in Windows.";
1890        const char *es1, *es2;
1891
1892        es2 = es2a;
1893        if (vol->vol_flags & VOLUME_IS_DIRTY)
1894            es1 = es1a;
1895        else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1896            es1 = es1b;
1897            es2 = es2b;
1898        } else {
1899            es1 = es1c;
1900            ntfs_warning(sb, "Unsupported volume flags 0x%x "
1901                    "encountered.",
1902                    (unsigned)le16_to_cpu(vol->vol_flags));
1903        }
1904        /* If a read-write mount, convert it to a read-only mount. */
1905        if (!(sb->s_flags & MS_RDONLY)) {
1906            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1907                    ON_ERRORS_CONTINUE))) {
1908                ntfs_error(sb, "%s and neither on_errors="
1909                        "continue nor on_errors="
1910                        "remount-ro was specified%s",
1911                        es1, es2);
1912                goto iput_vol_err_out;
1913            }
1914            sb->s_flags |= MS_RDONLY;
1915            ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1916        } else
1917            ntfs_warning(sb, "%s. Will not be able to remount "
1918                    "read-write%s", es1, es2);
1919        /*
1920         * Do not set NVolErrors() because ntfs_remount() re-checks the
1921         * flags which we need to do in case any flags have changed.
1922         */
1923    }
1924    /*
1925     * Get the inode for the logfile, check it and determine if the volume
1926     * was shutdown cleanly.
1927     */
1928    rp = NULL;
1929    if (!load_and_check_logfile(vol, &rp) ||
1930            !ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1931        static const char *es1a = "Failed to load $LogFile";
1932        static const char *es1b = "$LogFile is not clean";
1933        static const char *es2 = ". Mount in Windows.";
1934        const char *es1;
1935
1936        es1 = !vol->logfile_ino ? es1a : es1b;
1937        /* If a read-write mount, convert it to a read-only mount. */
1938        if (!(sb->s_flags & MS_RDONLY)) {
1939            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1940                    ON_ERRORS_CONTINUE))) {
1941                ntfs_error(sb, "%s and neither on_errors="
1942                        "continue nor on_errors="
1943                        "remount-ro was specified%s",
1944                        es1, es2);
1945                if (vol->logfile_ino) {
1946                    BUG_ON(!rp);
1947                    ntfs_free(rp);
1948                }
1949                goto iput_logfile_err_out;
1950            }
1951            sb->s_flags |= MS_RDONLY;
1952            ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1953        } else
1954            ntfs_warning(sb, "%s. Will not be able to remount "
1955                    "read-write%s", es1, es2);
1956        /* This will prevent a read-write remount. */
1957        NVolSetErrors(vol);
1958    }
1959    ntfs_free(rp);
1960#endif /* NTFS_RW */
1961    /* Get the root directory inode so we can do path lookups. */
1962    vol->root_ino = ntfs_iget(sb, FILE_root);
1963    if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1964        if (!IS_ERR(vol->root_ino))
1965            iput(vol->root_ino);
1966        ntfs_error(sb, "Failed to load root directory.");
1967        goto iput_logfile_err_out;
1968    }
1969#ifdef NTFS_RW
1970    /*
1971     * Check if Windows is suspended to disk on the target volume. If it
1972     * is hibernated, we must not write *anything* to the disk so set
1973     * NVolErrors() without setting the dirty volume flag and mount
1974     * read-only. This will prevent read-write remounting and it will also
1975     * prevent all writes.
1976     */
1977    err = check_windows_hibernation_status(vol);
1978    if (unlikely(err)) {
1979        static const char *es1a = "Failed to determine if Windows is "
1980                "hibernated";
1981        static const char *es1b = "Windows is hibernated";
1982        static const char *es2 = ". Run chkdsk.";
1983        const char *es1;
1984
1985        es1 = err < 0 ? es1a : es1b;
1986        /* If a read-write mount, convert it to a read-only mount. */
1987        if (!(sb->s_flags & MS_RDONLY)) {
1988            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1989                    ON_ERRORS_CONTINUE))) {
1990                ntfs_error(sb, "%s and neither on_errors="
1991                        "continue nor on_errors="
1992                        "remount-ro was specified%s",
1993                        es1, es2);
1994                goto iput_root_err_out;
1995            }
1996            sb->s_flags |= MS_RDONLY;
1997            ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
1998        } else
1999            ntfs_warning(sb, "%s. Will not be able to remount "
2000                    "read-write%s", es1, es2);
2001        /* This will prevent a read-write remount. */
2002        NVolSetErrors(vol);
2003    }
2004    /* If (still) a read-write mount, mark the volume dirty. */
2005    if (!(sb->s_flags & MS_RDONLY) &&
2006            ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
2007        static const char *es1 = "Failed to set dirty bit in volume "
2008                "information flags";
2009        static const char *es2 = ". Run chkdsk.";
2010
2011        /* Convert to a read-only mount. */
2012        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2013                ON_ERRORS_CONTINUE))) {
2014            ntfs_error(sb, "%s and neither on_errors=continue nor "
2015                    "on_errors=remount-ro was specified%s",
2016                    es1, es2);
2017            goto iput_root_err_out;
2018        }
2019        ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2020        sb->s_flags |= MS_RDONLY;
2021        /*
2022         * Do not set NVolErrors() because ntfs_remount() might manage
2023         * to set the dirty flag in which case all would be well.
2024         */
2025    }
2026#if 0
2027    // TODO: Enable this code once we start modifying anything that is
2028    // different between NTFS 1.2 and 3.x...
2029    /*
2030     * If (still) a read-write mount, set the NT4 compatibility flag on
2031     * newer NTFS version volumes.
2032     */
2033    if (!(sb->s_flags & MS_RDONLY) && (vol->major_ver > 1) &&
2034            ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
2035        static const char *es1 = "Failed to set NT4 compatibility flag";
2036        static const char *es2 = ". Run chkdsk.";
2037
2038        /* Convert to a read-only mount. */
2039        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2040                ON_ERRORS_CONTINUE))) {
2041            ntfs_error(sb, "%s and neither on_errors=continue nor "
2042                    "on_errors=remount-ro was specified%s",
2043                    es1, es2);
2044            goto iput_root_err_out;
2045        }
2046        ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2047        sb->s_flags |= MS_RDONLY;
2048        NVolSetErrors(vol);
2049    }
2050#endif
2051    /* If (still) a read-write mount, empty the logfile. */
2052    if (!(sb->s_flags & MS_RDONLY) &&
2053            !ntfs_empty_logfile(vol->logfile_ino)) {
2054        static const char *es1 = "Failed to empty $LogFile";
2055        static const char *es2 = ". Mount in Windows.";
2056
2057        /* Convert to a read-only mount. */
2058        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2059                ON_ERRORS_CONTINUE))) {
2060            ntfs_error(sb, "%s and neither on_errors=continue nor "
2061                    "on_errors=remount-ro was specified%s",
2062                    es1, es2);
2063            goto iput_root_err_out;
2064        }
2065        ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2066        sb->s_flags |= MS_RDONLY;
2067        NVolSetErrors(vol);
2068    }
2069#endif /* NTFS_RW */
2070    /* If on NTFS versions before 3.0, we are done. */
2071    if (unlikely(vol->major_ver < 3))
2072        return true;
2073    /* NTFS 3.0+ specific initialization. */
2074    /* Get the security descriptors inode. */
2075    vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2076    if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2077        if (!IS_ERR(vol->secure_ino))
2078            iput(vol->secure_ino);
2079        ntfs_error(sb, "Failed to load $Secure.");
2080        goto iput_root_err_out;
2081    }
2082    // TODO: Initialize security.
2083    /* Get the extended system files' directory inode. */
2084    vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2085    if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2086        if (!IS_ERR(vol->extend_ino))
2087            iput(vol->extend_ino);
2088        ntfs_error(sb, "Failed to load $Extend.");
2089        goto iput_sec_err_out;
2090    }
2091#ifdef NTFS_RW
2092    /* Find the quota file, load it if present, and set it up. */
2093    if (!load_and_init_quota(vol)) {
2094        static const char *es1 = "Failed to load $Quota";
2095        static const char *es2 = ". Run chkdsk.";
2096
2097        /* If a read-write mount, convert it to a read-only mount. */
2098        if (!(sb->s_flags & MS_RDONLY)) {
2099            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2100                    ON_ERRORS_CONTINUE))) {
2101                ntfs_error(sb, "%s and neither on_errors="
2102                        "continue nor on_errors="
2103                        "remount-ro was specified%s",
2104                        es1, es2);
2105                goto iput_quota_err_out;
2106            }
2107            sb->s_flags |= MS_RDONLY;
2108            ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2109        } else
2110            ntfs_warning(sb, "%s. Will not be able to remount "
2111                    "read-write%s", es1, es2);
2112        /* This will prevent a read-write remount. */
2113        NVolSetErrors(vol);
2114    }
2115    /* If (still) a read-write mount, mark the quotas out of date. */
2116    if (!(sb->s_flags & MS_RDONLY) &&
2117            !ntfs_mark_quotas_out_of_date(vol)) {
2118        static const char *es1 = "Failed to mark quotas out of date";
2119        static const char *es2 = ". Run chkdsk.";
2120
2121        /* Convert to a read-only mount. */
2122        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2123                ON_ERRORS_CONTINUE))) {
2124            ntfs_error(sb, "%s and neither on_errors=continue nor "
2125                    "on_errors=remount-ro was specified%s",
2126                    es1, es2);
2127            goto iput_quota_err_out;
2128        }
2129        ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2130        sb->s_flags |= MS_RDONLY;
2131        NVolSetErrors(vol);
2132    }
2133    /*
2134     * Find the transaction log file ($UsnJrnl), load it if present, check
2135     * it, and set it up.
2136     */
2137    if (!load_and_init_usnjrnl(vol)) {
2138        static const char *es1 = "Failed to load $UsnJrnl";
2139        static const char *es2 = ". Run chkdsk.";
2140
2141        /* If a read-write mount, convert it to a read-only mount. */
2142        if (!(sb->s_flags & MS_RDONLY)) {
2143            if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2144                    ON_ERRORS_CONTINUE))) {
2145                ntfs_error(sb, "%s and neither on_errors="
2146                        "continue nor on_errors="
2147                        "remount-ro was specified%s",
2148                        es1, es2);
2149                goto iput_usnjrnl_err_out;
2150            }
2151            sb->s_flags |= MS_RDONLY;
2152            ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2153        } else
2154            ntfs_warning(sb, "%s. Will not be able to remount "
2155                    "read-write%s", es1, es2);
2156        /* This will prevent a read-write remount. */
2157        NVolSetErrors(vol);
2158    }
2159    /* If (still) a read-write mount, stamp the transaction log. */
2160    if (!(sb->s_flags & MS_RDONLY) && !ntfs_stamp_usnjrnl(vol)) {
2161        static const char *es1 = "Failed to stamp transaction log "
2162                "($UsnJrnl)";
2163        static const char *es2 = ". Run chkdsk.";
2164
2165        /* Convert to a read-only mount. */
2166        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2167                ON_ERRORS_CONTINUE))) {
2168            ntfs_error(sb, "%s and neither on_errors=continue nor "
2169                    "on_errors=remount-ro was specified%s",
2170                    es1, es2);
2171            goto iput_usnjrnl_err_out;
2172        }
2173        ntfs_error(sb, "%s. Mounting read-only%s", es1, es2);
2174        sb->s_flags |= MS_RDONLY;
2175        NVolSetErrors(vol);
2176    }
2177#endif /* NTFS_RW */
2178    return true;
2179#ifdef NTFS_RW
2180iput_usnjrnl_err_out:
2181    if (vol->usnjrnl_j_ino)
2182        iput(vol->usnjrnl_j_ino);
2183    if (vol->usnjrnl_max_ino)
2184        iput(vol->usnjrnl_max_ino);
2185    if (vol->usnjrnl_ino)
2186        iput(vol->usnjrnl_ino);
2187iput_quota_err_out:
2188    if (vol->quota_q_ino)
2189        iput(vol->quota_q_ino);
2190    if (vol->quota_ino)
2191        iput(vol->quota_ino);
2192    iput(vol->extend_ino);
2193#endif /* NTFS_RW */
2194iput_sec_err_out:
2195    iput(vol->secure_ino);
2196iput_root_err_out:
2197    iput(vol->root_ino);
2198iput_logfile_err_out:
2199#ifdef NTFS_RW
2200    if (vol->logfile_ino)
2201        iput(vol->logfile_ino);
2202iput_vol_err_out:
2203#endif /* NTFS_RW */
2204    iput(vol->vol_ino);
2205iput_lcnbmp_err_out:
2206    iput(vol->lcnbmp_ino);
2207iput_attrdef_err_out:
2208    vol->attrdef_size = 0;
2209    if (vol->attrdef) {
2210        ntfs_free(vol->attrdef);
2211        vol->attrdef = NULL;
2212    }
2213#ifdef NTFS_RW
2214iput_upcase_err_out:
2215#endif /* NTFS_RW */
2216    vol->upcase_len = 0;
2217    mutex_lock(&ntfs_lock);
2218    if (vol->upcase == default_upcase) {
2219        ntfs_nr_upcase_users--;
2220        vol->upcase = NULL;
2221    }
2222    mutex_unlock(&ntfs_lock);
2223    if (vol->upcase) {
2224        ntfs_free(vol->upcase);
2225        vol->upcase = NULL;
2226    }
2227iput_mftbmp_err_out:
2228    iput(vol->mftbmp_ino);
2229iput_mirr_err_out:
2230#ifdef NTFS_RW
2231    if (vol->mftmirr_ino)
2232        iput(vol->mftmirr_ino);
2233#endif /* NTFS_RW */
2234    return false;
2235}
2236
2237/**
2238 * ntfs_put_super - called by the vfs to unmount a volume
2239 * @sb: vfs superblock of volume to unmount
2240 *
2241 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2242 * the volume is being unmounted (umount system call has been invoked) and it
2243 * releases all inodes and memory belonging to the NTFS specific part of the
2244 * super block.
2245 */
2246static void ntfs_put_super(struct super_block *sb)
2247{
2248    ntfs_volume *vol = NTFS_SB(sb);
2249
2250    ntfs_debug("Entering.");
2251
2252#ifdef NTFS_RW
2253    /*
2254     * Commit all inodes while they are still open in case some of them
2255     * cause others to be dirtied.
2256     */
2257    ntfs_commit_inode(vol->vol_ino);
2258
2259    /* NTFS 3.0+ specific. */
2260    if (vol->major_ver >= 3) {
2261        if (vol->usnjrnl_j_ino)
2262            ntfs_commit_inode(vol->usnjrnl_j_ino);
2263        if (vol->usnjrnl_max_ino)
2264            ntfs_commit_inode(vol->usnjrnl_max_ino);
2265        if (vol->usnjrnl_ino)
2266            ntfs_commit_inode(vol->usnjrnl_ino);
2267        if (vol->quota_q_ino)
2268            ntfs_commit_inode(vol->quota_q_ino);
2269        if (vol->quota_ino)
2270            ntfs_commit_inode(vol->quota_ino);
2271        if (vol->extend_ino)
2272            ntfs_commit_inode(vol->extend_ino);
2273        if (vol->secure_ino)
2274            ntfs_commit_inode(vol->secure_ino);
2275    }
2276
2277    ntfs_commit_inode(vol->root_ino);
2278
2279    down_write(&vol->lcnbmp_lock);
2280    ntfs_commit_inode(vol->lcnbmp_ino);
2281    up_write(&vol->lcnbmp_lock);
2282
2283    down_write(&vol->mftbmp_lock);
2284    ntfs_commit_inode(vol->mftbmp_ino);
2285    up_write(&vol->mftbmp_lock);
2286
2287    if (vol->logfile_ino)
2288        ntfs_commit_inode(vol->logfile_ino);
2289
2290    if (vol->mftmirr_ino)
2291        ntfs_commit_inode(vol->mftmirr_ino);
2292    ntfs_commit_inode(vol->mft_ino);
2293
2294    /*
2295     * If a read-write mount and no volume errors have occurred, mark the
2296     * volume clean. Also, re-commit all affected inodes.
2297     */
2298    if (!(sb->s_flags & MS_RDONLY)) {
2299        if (!NVolErrors(vol)) {
2300            if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2301                ntfs_warning(sb, "Failed to clear dirty bit "
2302                        "in volume information "
2303                        "flags. Run chkdsk.");
2304            ntfs_commit_inode(vol->vol_ino);
2305            ntfs_commit_inode(vol->root_ino);
2306            if (vol->mftmirr_ino)
2307                ntfs_commit_inode(vol->mftmirr_ino);
2308            ntfs_commit_inode(vol->mft_ino);
2309        } else {
2310            ntfs_warning(sb, "Volume has errors. Leaving volume "
2311                    "marked dirty. Run chkdsk.");
2312        }
2313    }
2314#endif /* NTFS_RW */
2315
2316    iput(vol->vol_ino);
2317    vol->vol_ino = NULL;
2318
2319    /* NTFS 3.0+ specific clean up. */
2320    if (vol->major_ver >= 3) {
2321#ifdef NTFS_RW
2322        if (vol->usnjrnl_j_ino) {
2323            iput(vol->usnjrnl_j_ino);
2324            vol->usnjrnl_j_ino = NULL;
2325        }
2326        if (vol->usnjrnl_max_ino) {
2327            iput(vol->usnjrnl_max_ino);
2328            vol->usnjrnl_max_ino = NULL;
2329        }
2330        if (vol->usnjrnl_ino) {
2331            iput(vol->usnjrnl_ino);
2332            vol->usnjrnl_ino = NULL;
2333        }
2334        if (vol->quota_q_ino) {
2335            iput(vol->quota_q_ino);
2336            vol->quota_q_ino = NULL;
2337        }
2338        if (vol->quota_ino) {
2339            iput(vol->quota_ino);
2340            vol->quota_ino = NULL;
2341        }
2342#endif /* NTFS_RW */
2343        if (vol->extend_ino) {
2344            iput(vol->extend_ino);
2345            vol->extend_ino = NULL;
2346        }
2347        if (vol->secure_ino) {
2348            iput(vol->secure_ino);
2349            vol->secure_ino = NULL;
2350        }
2351    }
2352
2353    iput(vol->root_ino);
2354    vol->root_ino = NULL;
2355
2356    down_write(&vol->lcnbmp_lock);
2357    iput(vol->lcnbmp_ino);
2358    vol->lcnbmp_ino = NULL;
2359    up_write(&vol->lcnbmp_lock);
2360
2361    down_write(&vol->mftbmp_lock);
2362    iput(vol->mftbmp_ino);
2363    vol->mftbmp_ino = NULL;
2364    up_write(&vol->mftbmp_lock);
2365
2366#ifdef NTFS_RW
2367    if (vol->logfile_ino) {
2368        iput(vol->logfile_ino);
2369        vol->logfile_ino = NULL;
2370    }
2371    if (vol->mftmirr_ino) {
2372        /* Re-commit the mft mirror and mft just in case. */
2373        ntfs_commit_inode(vol->mftmirr_ino);
2374        ntfs_commit_inode(vol->mft_ino);
2375        iput(vol->mftmirr_ino);
2376        vol->mftmirr_ino = NULL;
2377    }
2378    /*
2379     * We should have no dirty inodes left, due to
2380     * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2381     * the underlying mft records are written out and cleaned.
2382     */
2383    ntfs_commit_inode(vol->mft_ino);
2384    write_inode_now(vol->mft_ino, 1);
2385#endif /* NTFS_RW */
2386
2387    iput(vol->mft_ino);
2388    vol->mft_ino = NULL;
2389
2390    /* Throw away the table of attribute definitions. */
2391    vol->attrdef_size = 0;
2392    if (vol->attrdef) {
2393        ntfs_free(vol->attrdef);
2394        vol->attrdef = NULL;
2395    }
2396    vol->upcase_len = 0;
2397    /*
2398     * Destroy the global default upcase table if necessary. Also decrease
2399     * the number of upcase users if we are a user.
2400     */
2401    mutex_lock(&ntfs_lock);
2402    if (vol->upcase == default_upcase) {
2403        ntfs_nr_upcase_users--;
2404        vol->upcase = NULL;
2405    }
2406    if (!ntfs_nr_upcase_users && default_upcase) {
2407        ntfs_free(default_upcase);
2408        default_upcase = NULL;
2409    }
2410    if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2411        free_compression_buffers();
2412    mutex_unlock(&ntfs_lock);
2413    if (vol->upcase) {
2414        ntfs_free(vol->upcase);
2415        vol->upcase = NULL;
2416    }
2417
2418    unload_nls(vol->nls_map);
2419
2420    sb->s_fs_info = NULL;
2421    kfree(vol);
2422}
2423
2424/**
2425 * get_nr_free_clusters - return the number of free clusters on a volume
2426 * @vol: ntfs volume for which to obtain free cluster count
2427 *
2428 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2429 * actually calculate the number of clusters in use instead because this
2430 * allows us to not care about partial pages as these will be just zero filled
2431 * and hence not be counted as allocated clusters.
2432 *
2433 * The only particularity is that clusters beyond the end of the logical ntfs
2434 * volume will be marked as allocated to prevent errors which means we have to
2435 * discount those at the end. This is important as the cluster bitmap always
2436 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2437 * the logical volume and marked in use when they are not as they do not exist.
2438 *
2439 * If any pages cannot be read we assume all clusters in the erroring pages are
2440 * in use. This means we return an underestimate on errors which is better than
2441 * an overestimate.
2442 */
2443static s64 get_nr_free_clusters(ntfs_volume *vol)
2444{
2445    s64 nr_free = vol->nr_clusters;
2446    struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2447    struct page *page;
2448    pgoff_t index, max_index;
2449
2450    ntfs_debug("Entering.");
2451    /* Serialize accesses to the cluster bitmap. */
2452    down_read(&vol->lcnbmp_lock);
2453    /*
2454     * Convert the number of bits into bytes rounded up, then convert into
2455     * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2456     * full and one partial page max_index = 2.
2457     */
2458    max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2459            PAGE_CACHE_SHIFT;
2460    /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2461    ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2462            max_index, PAGE_CACHE_SIZE / 4);
2463    for (index = 0; index < max_index; index++) {
2464        unsigned long *kaddr;
2465
2466        /*
2467         * Read the page from page cache, getting it from backing store
2468         * if necessary, and increment the use count.
2469         */
2470        page = read_mapping_page(mapping, index, NULL);
2471        /* Ignore pages which errored synchronously. */
2472        if (IS_ERR(page)) {
2473            ntfs_debug("read_mapping_page() error. Skipping "
2474                    "page (index 0x%lx).", index);
2475            nr_free -= PAGE_CACHE_SIZE * 8;
2476            continue;
2477        }
2478        kaddr = kmap_atomic(page, KM_USER0);
2479        /*
2480         * Subtract the number of set bits. If this
2481         * is the last page and it is partial we don't really care as
2482         * it just means we do a little extra work but it won't affect
2483         * the result as all out of range bytes are set to zero by
2484         * ntfs_readpage().
2485         */
2486        nr_free -= bitmap_weight(kaddr,
2487                    PAGE_CACHE_SIZE * BITS_PER_BYTE);
2488        kunmap_atomic(kaddr, KM_USER0);
2489        page_cache_release(page);
2490    }
2491    ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2492    /*
2493     * Fixup for eventual bits outside logical ntfs volume (see function
2494     * description above).
2495     */
2496    if (vol->nr_clusters & 63)
2497        nr_free += 64 - (vol->nr_clusters & 63);
2498    up_read(&vol->lcnbmp_lock);
2499    /* If errors occurred we may well have gone below zero, fix this. */
2500    if (nr_free < 0)
2501        nr_free = 0;
2502    ntfs_debug("Exiting.");
2503    return nr_free;
2504}
2505
2506/**
2507 * __get_nr_free_mft_records - return the number of free inodes on a volume
2508 * @vol: ntfs volume for which to obtain free inode count
2509 * @nr_free: number of mft records in filesystem
2510 * @max_index: maximum number of pages containing set bits
2511 *
2512 * Calculate the number of free mft records (inodes) on the mounted NTFS
2513 * volume @vol. We actually calculate the number of mft records in use instead
2514 * because this allows us to not care about partial pages as these will be just
2515 * zero filled and hence not be counted as allocated mft record.
2516 *
2517 * If any pages cannot be read we assume all mft records in the erroring pages
2518 * are in use. This means we return an underestimate on errors which is better
2519 * than an overestimate.
2520 *
2521 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2522 */
2523static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2524        s64 nr_free, const pgoff_t max_index)
2525{
2526    struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2527    struct page *page;
2528    pgoff_t index;
2529
2530    ntfs_debug("Entering.");
2531    /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2532    ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2533            "0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2534    for (index = 0; index < max_index; index++) {
2535        unsigned long *kaddr;
2536
2537        /*
2538         * Read the page from page cache, getting it from backing store
2539         * if necessary, and increment the use count.
2540         */
2541        page = read_mapping_page(mapping, index, NULL);
2542        /* Ignore pages which errored synchronously. */
2543        if (IS_ERR(page)) {
2544            ntfs_debug("read_mapping_page() error. Skipping "
2545                    "page (index 0x%lx).", index);
2546            nr_free -= PAGE_CACHE_SIZE * 8;
2547            continue;
2548        }
2549        kaddr = kmap_atomic(page, KM_USER0);
2550        /*
2551         * Subtract the number of set bits. If this
2552         * is the last page and it is partial we don't really care as
2553         * it just means we do a little extra work but it won't affect
2554         * the result as all out of range bytes are set to zero by
2555         * ntfs_readpage().
2556         */
2557        nr_free -= bitmap_weight(kaddr,
2558                    PAGE_CACHE_SIZE * BITS_PER_BYTE);
2559        kunmap_atomic(kaddr, KM_USER0);
2560        page_cache_release(page);
2561    }
2562    ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2563            index - 1);
2564    /* If errors occurred we may well have gone below zero, fix this. */
2565    if (nr_free < 0)
2566        nr_free = 0;
2567    ntfs_debug("Exiting.");
2568    return nr_free;
2569}
2570
2571/**
2572 * ntfs_statfs - return information about mounted NTFS volume
2573 * @dentry: dentry from mounted volume
2574 * @sfs: statfs structure in which to return the information
2575 *
2576 * Return information about the mounted NTFS volume @dentry in the statfs structure
2577 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2578 * called). We interpret the values to be correct of the moment in time at
2579 * which we are called. Most values are variable otherwise and this isn't just
2580 * the free values but the totals as well. For example we can increase the
2581 * total number of file nodes if we run out and we can keep doing this until
2582 * there is no more space on the volume left at all.
2583 *
2584 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2585 * ustat system calls.
2586 *
2587 * Return 0 on success or -errno on error.
2588 */
2589static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2590{
2591    struct super_block *sb = dentry->d_sb;
2592    s64 size;
2593    ntfs_volume *vol = NTFS_SB(sb);
2594    ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2595    pgoff_t max_index;
2596    unsigned long flags;
2597
2598    ntfs_debug("Entering.");
2599    /* Type of filesystem. */
2600    sfs->f_type = NTFS_SB_MAGIC;
2601    /* Optimal transfer block size. */
2602    sfs->f_bsize = PAGE_CACHE_SIZE;
2603    /*
2604     * Total data blocks in filesystem in units of f_bsize and since
2605     * inodes are also stored in data blocs ($MFT is a file) this is just
2606     * the total clusters.
2607     */
2608    sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2609                PAGE_CACHE_SHIFT;
2610    /* Free data blocks in filesystem in units of f_bsize. */
2611    size = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2612                PAGE_CACHE_SHIFT;
2613    if (size < 0LL)
2614        size = 0LL;
2615    /* Free blocks avail to non-superuser, same as above on NTFS. */
2616    sfs->f_bavail = sfs->f_bfree = size;
2617    /* Serialize accesses to the inode bitmap. */
2618    down_read(&vol->mftbmp_lock);
2619    read_lock_irqsave(&mft_ni->size_lock, flags);
2620    size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2621    /*
2622     * Convert the maximum number of set bits into bytes rounded up, then
2623     * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2624     * have one full and one partial page max_index = 2.
2625     */
2626    max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2627            + 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2628    read_unlock_irqrestore(&mft_ni->size_lock, flags);
2629    /* Number of inodes in filesystem (at this point in time). */
2630    sfs->f_files = size;
2631    /* Free inodes in fs (based on current total count). */
2632    sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2633    up_read(&vol->mftbmp_lock);
2634    /*
2635     * File system id. This is extremely *nix flavour dependent and even
2636     * within Linux itself all fs do their own thing. I interpret this to
2637     * mean a unique id associated with the mounted fs and not the id
2638     * associated with the filesystem driver, the latter is already given
2639     * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2640     * volume serial number splitting it into two 32-bit parts. We enter
2641     * the least significant 32-bits in f_fsid[0] and the most significant
2642     * 32-bits in f_fsid[1].
2643     */
2644    sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2645    sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2646    /* Maximum length of filenames. */
2647    sfs->f_namelen = NTFS_MAX_NAME_LEN;
2648    return 0;
2649}
2650
2651#ifdef NTFS_RW
2652static int ntfs_write_inode(struct inode *vi, struct writeback_control *wbc)
2653{
2654    return __ntfs_write_inode(vi, wbc->sync_mode == WB_SYNC_ALL);
2655}
2656#endif
2657
2658/**
2659 * The complete super operations.
2660 */
2661static const struct super_operations ntfs_sops = {
2662    .alloc_inode = ntfs_alloc_big_inode, /* VFS: Allocate new inode. */
2663    .destroy_inode = ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2664#ifdef NTFS_RW
2665    //.dirty_inode = NULL, /* VFS: Called from
2666    // __mark_inode_dirty(). */
2667    .write_inode = ntfs_write_inode, /* VFS: Write dirty inode to
2668                           disk. */
2669    //.drop_inode = NULL, /* VFS: Called just after the
2670    // inode reference count has
2671    // been decreased to zero.
2672    // NOTE: The inode lock is
2673    // held. See fs/inode.c::
2674    // generic_drop_inode(). */
2675    //.delete_inode = NULL, /* VFS: Delete inode from disk.
2676    // Called when i_count becomes
2677    // 0 and i_nlink is also 0. */
2678    //.write_super = NULL, /* Flush dirty super block to
2679    // disk. */
2680    //.sync_fs = NULL, /* ? */
2681    //.write_super_lockfs = NULL, /* ? */
2682    //.unlockfs = NULL, /* ? */
2683#endif /* NTFS_RW */
2684    .put_super = ntfs_put_super, /* Syscall: umount. */
2685    .statfs = ntfs_statfs, /* Syscall: statfs */
2686    .remount_fs = ntfs_remount, /* Syscall: mount -o remount. */
2687    .evict_inode = ntfs_evict_big_inode, /* VFS: Called when an inode is
2688                           removed from memory. */
2689    //.umount_begin = NULL, /* Forced umount. */
2690    .show_options = ntfs_show_options, /* Show mount options in
2691                           proc. */
2692};
2693
2694/**
2695 * ntfs_fill_super - mount an ntfs filesystem
2696 * @sb: super block of ntfs filesystem to mount
2697 * @opt: string containing the mount options
2698 * @silent: silence error output
2699 *
2700 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2701 * with the mount otions in @data with the NTFS filesystem.
2702 *
2703 * If @silent is true, remain silent even if errors are detected. This is used
2704 * during bootup, when the kernel tries to mount the root filesystem with all
2705 * registered filesystems one after the other until one succeeds. This implies
2706 * that all filesystems except the correct one will quite correctly and
2707 * expectedly return an error, but nobody wants to see error messages when in
2708 * fact this is what is supposed to happen.
2709 *
2710 * NOTE: @sb->s_flags contains the mount options flags.
2711 */
2712static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2713{
2714    ntfs_volume *vol;
2715    struct buffer_head *bh;
2716    struct inode *tmp_ino;
2717    int blocksize, result;
2718
2719    /*
2720     * We do a pretty difficult piece of bootstrap by reading the
2721     * MFT (and other metadata) from disk into memory. We'll only
2722     * release this metadata during umount, so the locking patterns
2723     * observed during bootstrap do not count. So turn off the
2724     * observation of locking patterns (strictly for this context
2725     * only) while mounting NTFS. [The validator is still active
2726     * otherwise, even for this context: it will for example record
2727     * lock class registrations.]
2728     */
2729    lockdep_off();
2730    ntfs_debug("Entering.");
2731#ifndef NTFS_RW
2732    sb->s_flags |= MS_RDONLY;
2733#endif /* ! NTFS_RW */
2734    /* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2735    sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2736    vol = NTFS_SB(sb);
2737    if (!vol) {
2738        if (!silent)
2739            ntfs_error(sb, "Allocation of NTFS volume structure "
2740                    "failed. Aborting mount...");
2741        lockdep_on();
2742        return -ENOMEM;
2743    }
2744    /* Initialize ntfs_volume structure. */
2745    *vol = (ntfs_volume) {
2746        .sb = sb,
2747        /*
2748         * Default is group and other don't have any access to files or
2749         * directories while owner has full access. Further, files by
2750         * default are not executable but directories are of course
2751         * browseable.
2752         */
2753        .fmask = 0177,
2754        .dmask = 0077,
2755    };
2756    init_rwsem(&vol->mftbmp_lock);
2757    init_rwsem(&vol->lcnbmp_lock);
2758
2759    /* By default, enable sparse support. */
2760    NVolSetSparseEnabled(vol);
2761
2762    /* Important to get the mount options dealt with now. */
2763    if (!parse_options(vol, (char*)opt))
2764        goto err_out_now;
2765
2766    /* We support sector sizes up to the PAGE_CACHE_SIZE. */
2767    if (bdev_logical_block_size(sb->s_bdev) > PAGE_CACHE_SIZE) {
2768        if (!silent)
2769            ntfs_error(sb, "Device has unsupported sector size "
2770                    "(%i). The maximum supported sector "
2771                    "size on this architecture is %lu "
2772                    "bytes.",
2773                    bdev_logical_block_size(sb->s_bdev),
2774                    PAGE_CACHE_SIZE);
2775        goto err_out_now;
2776    }
2777    /*
2778     * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2779     * sector size, whichever is bigger.
2780     */
2781    blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2782    if (blocksize < NTFS_BLOCK_SIZE) {
2783        if (!silent)
2784            ntfs_error(sb, "Unable to set device block size.");
2785        goto err_out_now;
2786    }
2787    BUG_ON(blocksize != sb->s_blocksize);
2788    ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2789            blocksize, sb->s_blocksize_bits);
2790    /* Determine the size of the device in units of block_size bytes. */
2791    if (!i_size_read(sb->s_bdev->bd_inode)) {
2792        if (!silent)
2793            ntfs_error(sb, "Unable to determine device size.");
2794        goto err_out_now;
2795    }
2796    vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2797            sb->s_blocksize_bits;
2798    /* Read the boot sector and return unlocked buffer head to it. */
2799    if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2800        if (!silent)
2801            ntfs_error(sb, "Not an NTFS volume.");
2802        goto err_out_now;
2803    }
2804    /*
2805     * Extract the data from the boot sector and setup the ntfs volume
2806     * using it.
2807     */
2808    result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2809    brelse(bh);
2810    if (!result) {
2811        if (!silent)
2812            ntfs_error(sb, "Unsupported NTFS filesystem.");
2813        goto err_out_now;
2814    }
2815    /*
2816     * If the boot sector indicates a sector size bigger than the current
2817     * device block size, switch the device block size to the sector size.
2818     * TODO: It may be possible to support this case even when the set
2819     * below fails, we would just be breaking up the i/o for each sector
2820     * into multiple blocks for i/o purposes but otherwise it should just
2821     * work. However it is safer to leave disabled until someone hits this
2822     * error message and then we can get them to try it without the setting
2823     * so we know for sure that it works.
2824     */
2825    if (vol->sector_size > blocksize) {
2826        blocksize = sb_set_blocksize(sb, vol->sector_size);
2827        if (blocksize != vol->sector_size) {
2828            if (!silent)
2829                ntfs_error(sb, "Unable to set device block "
2830                        "size to sector size (%i).",
2831                        vol->sector_size);
2832            goto err_out_now;
2833        }
2834        BUG_ON(blocksize != sb->s_blocksize);
2835        vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2836                sb->s_blocksize_bits;
2837        ntfs_debug("Changed device block size to %i bytes (block size "
2838                "bits %i) to match volume sector size.",
2839                blocksize, sb->s_blocksize_bits);
2840    }
2841    /* Initialize the cluster and mft allocators. */
2842    ntfs_setup_allocators(vol);
2843    /* Setup remaining fields in the super block. */
2844    sb->s_magic = NTFS_SB_MAGIC;
2845    /*
2846     * Ntfs allows 63 bits for the file size, i.e. correct would be:
2847     * sb->s_maxbytes = ~0ULL >> 1;
2848     * But the kernel uses a long as the page cache page index which on
2849     * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2850     * defined to the maximum the page cache page index can cope with
2851     * without overflowing the index or to 2^63 - 1, whichever is smaller.
2852     */
2853    sb->s_maxbytes = MAX_LFS_FILESIZE;
2854    /* Ntfs measures time in 100ns intervals. */
2855    sb->s_time_gran = 100;
2856    /*
2857     * Now load the metadata required for the page cache and our address
2858     * space operations to function. We do this by setting up a specialised
2859     * read_inode method and then just calling the normal iget() to obtain
2860     * the inode for $MFT which is sufficient to allow our normal inode
2861     * operations and associated address space operations to function.
2862     */
2863    sb->s_op = &ntfs_sops;
2864    tmp_ino = new_inode(sb);
2865    if (!tmp_ino) {
2866        if (!silent)
2867            ntfs_error(sb, "Failed to load essential metadata.");
2868        goto err_out_now;
2869    }
2870    tmp_ino->i_ino = FILE_MFT;
2871    insert_inode_hash(tmp_ino);
2872    if (ntfs_read_inode_mount(tmp_ino) < 0) {
2873        if (!silent)
2874            ntfs_error(sb, "Failed to load essential metadata.");
2875        goto iput_tmp_ino_err_out_now;
2876    }
2877    mutex_lock(&ntfs_lock);
2878    /*
2879     * The current mount is a compression user if the cluster size is
2880     * less than or equal 4kiB.
2881     */
2882    if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2883        result = allocate_compression_buffers();
2884        if (result) {
2885            ntfs_error(NULL, "Failed to allocate buffers "
2886                    "for compression engine.");
2887            ntfs_nr_compression_users--;
2888            mutex_unlock(&ntfs_lock);
2889            goto iput_tmp_ino_err_out_now;
2890        }
2891    }
2892    /*
2893     * Generate the global default upcase table if necessary. Also
2894     * temporarily increment the number of upcase users to avoid race
2895     * conditions with concurrent (u)mounts.
2896     */
2897    if (!default_upcase)
2898        default_upcase = generate_default_upcase();
2899    ntfs_nr_upcase_users++;
2900    mutex_unlock(&ntfs_lock);
2901    /*
2902     * From now on, ignore @silent parameter. If we fail below this line,
2903     * it will be due to a corrupt fs or a system error, so we report it.
2904     */
2905    /*
2906     * Open the system files with normal access functions and complete
2907     * setting up the ntfs super block.
2908     */
2909    if (!load_system_files(vol)) {
2910        ntfs_error(sb, "Failed to load system files.");
2911        goto unl_upcase_iput_tmp_ino_err_out_now;
2912    }
2913    if ((sb->s_root = d_alloc_root(vol->root_ino))) {
2914        /* We grab a reference, simulating an ntfs_iget(). */
2915        ihold(vol->root_ino);
2916        ntfs_debug("Exiting, status successful.");
2917        /* Release the default upcase if it has no users. */
2918        mutex_lock(&ntfs_lock);
2919        if (!--ntfs_nr_upcase_users && default_upcase) {
2920            ntfs_free(default_upcase);
2921            default_upcase = NULL;
2922        }
2923        mutex_unlock(&ntfs_lock);
2924        sb->s_export_op = &ntfs_export_ops;
2925        lockdep_on();
2926        return 0;
2927    }
2928    ntfs_error(sb, "Failed to allocate root directory.");
2929    /* Clean up after the successful load_system_files() call from above. */
2930    // TODO: Use ntfs_put_super() instead of repeating all this code...
2931    // FIXME: Should mark the volume clean as the error is most likely
2932    // -ENOMEM.
2933    iput(vol->vol_ino);
2934    vol->vol_ino = NULL;
2935    /* NTFS 3.0+ specific clean up. */
2936    if (vol->major_ver >= 3) {
2937#ifdef NTFS_RW
2938        if (vol->usnjrnl_j_ino) {
2939            iput(vol->usnjrnl_j_ino);
2940            vol->usnjrnl_j_ino = NULL;
2941        }
2942        if (vol->usnjrnl_max_ino) {
2943            iput(vol->usnjrnl_max_ino);
2944            vol->usnjrnl_max_ino = NULL;
2945        }
2946        if (vol->usnjrnl_ino) {
2947            iput(vol->usnjrnl_ino);
2948            vol->usnjrnl_ino = NULL;
2949        }
2950        if (vol->quota_q_ino) {
2951            iput(vol->quota_q_ino);
2952            vol->quota_q_ino = NULL;
2953        }
2954        if (vol->quota_ino) {
2955            iput(vol->quota_ino);
2956            vol->quota_ino = NULL;
2957        }
2958#endif /* NTFS_RW */
2959        if (vol->extend_ino) {
2960            iput(vol->extend_ino);
2961            vol->extend_ino = NULL;
2962        }
2963        if (vol->secure_ino) {
2964            iput(vol->secure_ino);
2965            vol->secure_ino = NULL;
2966        }
2967    }
2968    iput(vol->root_ino);
2969    vol->root_ino = NULL;
2970    iput(vol->lcnbmp_ino);
2971    vol->lcnbmp_ino = NULL;
2972    iput(vol->mftbmp_ino);
2973    vol->mftbmp_ino = NULL;
2974#ifdef NTFS_RW
2975    if (vol->logfile_ino) {
2976        iput(vol->logfile_ino);
2977        vol->logfile_ino = NULL;
2978    }
2979    if (vol->mftmirr_ino) {
2980        iput(vol->mftmirr_ino);
2981        vol->mftmirr_ino = NULL;
2982    }
2983#endif /* NTFS_RW */
2984    /* Throw away the table of attribute definitions. */
2985    vol->attrdef_size = 0;
2986    if (vol->attrdef) {
2987        ntfs_free(vol->attrdef);
2988        vol->attrdef = NULL;
2989    }
2990    vol->upcase_len = 0;
2991    mutex_lock(&ntfs_lock);
2992    if (vol->upcase == default_upcase) {
2993        ntfs_nr_upcase_users--;
2994        vol->upcase = NULL;
2995    }
2996    mutex_unlock(&ntfs_lock);
2997    if (vol->upcase) {
2998        ntfs_free(vol->upcase);
2999        vol->upcase = NULL;
3000    }
3001    if (vol->nls_map) {
3002        unload_nls(vol->nls_map);
3003        vol->nls_map = NULL;
3004    }
3005    /* Error exit code path. */
3006unl_upcase_iput_tmp_ino_err_out_now:
3007    /*
3008     * Decrease the number of upcase users and destroy the global default
3009     * upcase table if necessary.
3010     */
3011    mutex_lock(&ntfs_lock);
3012    if (!--ntfs_nr_upcase_users && default_upcase) {
3013        ntfs_free(default_upcase);
3014        default_upcase = NULL;
3015    }
3016    if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
3017        free_compression_buffers();
3018    mutex_unlock(&ntfs_lock);
3019iput_tmp_ino_err_out_now:
3020    iput(tmp_ino);
3021    if (vol->mft_ino && vol->mft_ino != tmp_ino)
3022        iput(vol->mft_ino);
3023    vol->mft_ino = NULL;
3024    /* Errors at this stage are irrelevant. */
3025err_out_now:
3026    sb->s_fs_info = NULL;
3027    kfree(vol);
3028    ntfs_debug("Failed, returning -EINVAL.");
3029    lockdep_on();
3030    return -EINVAL;
3031}
3032
3033/*
3034 * This is a slab cache to optimize allocations and deallocations of Unicode
3035 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3036 * (255) Unicode characters + a terminating NULL Unicode character.
3037 */
3038struct kmem_cache *ntfs_name_cache;
3039
3040/* Slab caches for efficient allocation/deallocation of inodes. */
3041struct kmem_cache *ntfs_inode_cache;
3042struct kmem_cache *ntfs_big_inode_cache;
3043
3044/* Init once constructor for the inode slab cache. */
3045static void ntfs_big_inode_init_once(void *foo)
3046{
3047    ntfs_inode *ni = (ntfs_inode *)foo;
3048
3049    inode_init_once(VFS_I(ni));
3050}
3051
3052/*
3053 * Slab caches to optimize allocations and deallocations of attribute search
3054 * contexts and index contexts, respectively.
3055 */
3056struct kmem_cache *ntfs_attr_ctx_cache;
3057struct kmem_cache *ntfs_index_ctx_cache;
3058
3059/* Driver wide mutex. */
3060DEFINE_MUTEX(ntfs_lock);
3061
3062static struct dentry *ntfs_mount(struct file_system_type *fs_type,
3063    int flags, const char *dev_name, void *data)
3064{
3065    return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
3066}
3067
3068static struct file_system_type ntfs_fs_type = {
3069    .owner = THIS_MODULE,
3070    .name = "ntfs",
3071    .mount = ntfs_mount,
3072    .kill_sb = kill_block_super,
3073    .fs_flags = FS_REQUIRES_DEV,
3074};
3075
3076/* Stable names for the slab caches. */
3077static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3078static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3079static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3080static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3081static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3082
3083static int __init init_ntfs_fs(void)
3084{
3085    int err = 0;
3086
3087    /* This may be ugly but it results in pretty output so who cares. (-8 */
3088    printk(KERN_INFO "NTFS driver " NTFS_VERSION " [Flags: R/"
3089#ifdef NTFS_RW
3090            "W"
3091#else
3092            "O"
3093#endif
3094#ifdef DEBUG
3095            " DEBUG"
3096#endif
3097#ifdef MODULE
3098            " MODULE"
3099#endif
3100            "].\n");
3101
3102    ntfs_debug("Debug messages are enabled.");
3103
3104    ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3105            sizeof(ntfs_index_context), 0 /* offset */,
3106            SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3107    if (!ntfs_index_ctx_cache) {
3108        printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3109                ntfs_index_ctx_cache_name);
3110        goto ictx_err_out;
3111    }
3112    ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3113            sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3114            SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3115    if (!ntfs_attr_ctx_cache) {
3116        printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3117                ntfs_attr_ctx_cache_name);
3118        goto actx_err_out;
3119    }
3120
3121    ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3122            (NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3123            SLAB_HWCACHE_ALIGN, NULL);
3124    if (!ntfs_name_cache) {
3125        printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3126                ntfs_name_cache_name);
3127        goto name_err_out;
3128    }
3129
3130    ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3131            sizeof(ntfs_inode), 0,
3132            SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
3133    if (!ntfs_inode_cache) {
3134        printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3135                ntfs_inode_cache_name);
3136        goto inode_err_out;
3137    }
3138
3139    ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3140            sizeof(big_ntfs_inode), 0,
3141            SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
3142            ntfs_big_inode_init_once);
3143    if (!ntfs_big_inode_cache) {
3144        printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3145                ntfs_big_inode_cache_name);
3146        goto big_inode_err_out;
3147    }
3148
3149    /* Register the ntfs sysctls. */
3150    err = ntfs_sysctl(1);
3151    if (err) {
3152        printk(KERN_CRIT "NTFS: Failed to register NTFS sysctls!\n");
3153        goto sysctl_err_out;
3154    }
3155
3156    err = register_filesystem(&ntfs_fs_type);
3157    if (!err) {
3158        ntfs_debug("NTFS driver registered successfully.");
3159        return 0; /* Success! */
3160    }
3161    printk(KERN_CRIT "NTFS: Failed to register NTFS filesystem driver!\n");
3162
3163sysctl_err_out:
3164    kmem_cache_destroy(ntfs_big_inode_cache);
3165big_inode_err_out:
3166    kmem_cache_destroy(ntfs_inode_cache);
3167inode_err_out:
3168    kmem_cache_destroy(ntfs_name_cache);
3169name_err_out:
3170    kmem_cache_destroy(ntfs_attr_ctx_cache);
3171actx_err_out:
3172    kmem_cache_destroy(ntfs_index_ctx_cache);
3173ictx_err_out:
3174    if (!err) {
3175        printk(KERN_CRIT "NTFS: Aborting NTFS filesystem driver "
3176                "registration...\n");
3177        err = -ENOMEM;
3178    }
3179    return err;
3180}
3181
3182static void __exit exit_ntfs_fs(void)
3183{
3184    ntfs_debug("Unregistering NTFS driver.");
3185
3186    unregister_filesystem(&ntfs_fs_type);
3187    kmem_cache_destroy(ntfs_big_inode_cache);
3188    kmem_cache_destroy(ntfs_inode_cache);
3189    kmem_cache_destroy(ntfs_name_cache);
3190    kmem_cache_destroy(ntfs_attr_ctx_cache);
3191    kmem_cache_destroy(ntfs_index_ctx_cache);
3192    /* Unregister the ntfs sysctls. */
3193    ntfs_sysctl(0);
3194}
3195
3196MODULE_AUTHOR("Anton Altaparmakov <anton@tuxera.com>");
3197MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2011 Anton Altaparmakov and Tuxera Inc.");
3198MODULE_VERSION(NTFS_VERSION);
3199MODULE_LICENSE("GPL");
3200#ifdef DEBUG
3201module_param(debug_msgs, bool, 0);
3202MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3203#endif
3204
3205module_init(init_ntfs_fs)
3206module_exit(exit_ntfs_fs)
3207

Archive Download this file



interactive