linux/fs/ntfs/super.c
<<
>>
Prefs
   1/*
   2 * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
   3 *
   4 * Copyright (c) 2001-2007 Anton Altaparmakov
   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/smp_lock.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 "aops.h"
  43#include "layout.h"
  44#include "malloc.h"
  45#include "ntfs.h"
  46
  47/* Number of mounted filesystems which have compression enabled. */
  48static unsigned long ntfs_nr_compression_users;
  49
  50/* A global default upcase table and a corresponding reference count. */
  51static ntfschar *default_upcase = NULL;
  52static unsigned long ntfs_nr_upcase_users = 0;
  53
  54/* Error constants/strings used in inode.c::ntfs_show_options(). */
  55typedef enum {
  56        /* One of these must be present, default is ON_ERRORS_CONTINUE. */
  57        ON_ERRORS_PANIC                 = 0x01,
  58        ON_ERRORS_REMOUNT_RO            = 0x02,
  59        ON_ERRORS_CONTINUE              = 0x04,
  60        /* Optional, can be combined with any of the above. */
  61        ON_ERRORS_RECOVER               = 0x10,
  62} ON_ERRORS_ACTIONS;
  63
  64const option_t on_errors_arr[] = {
  65        { ON_ERRORS_PANIC,      "panic" },
  66        { ON_ERRORS_REMOUNT_RO, "remount-ro", },
  67        { ON_ERRORS_CONTINUE,   "continue", },
  68        { ON_ERRORS_RECOVER,    "recover" },
  69        { 0,                    NULL }
  70};
  71
  72/**
  73 * simple_getbool -
  74 *
  75 * Copied from old ntfs driver (which copied from vfat driver).
  76 */
  77static int simple_getbool(char *s, bool *setval)
  78{
  79        if (s) {
  80                if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
  81                        *setval = true;
  82                else if (!strcmp(s, "0") || !strcmp(s, "no") ||
  83                                                        !strcmp(s, "false"))
  84                        *setval = false;
  85                else
  86                        return 0;
  87        } else
  88                *setval = true;
  89        return 1;
  90}
  91
  92/**
  93 * parse_options - parse the (re)mount options
  94 * @vol:        ntfs volume
  95 * @opt:        string containing the (re)mount options
  96 *
  97 * Parse the recognized options in @opt for the ntfs volume described by @vol.
  98 */
  99static bool parse_options(ntfs_volume *vol, char *opt)
 100{
 101        char *p, *v, *ov;
 102        static char *utf8 = "utf8";
 103        int errors = 0, sloppy = 0;
 104        uid_t uid = (uid_t)-1;
 105        gid_t gid = (gid_t)-1;
 106        mode_t fmask = (mode_t)-1, dmask = (mode_t)-1;
 107        int mft_zone_multiplier = -1, on_errors = -1;
 108        int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
 109        struct nls_table *nls_map = NULL, *old_nls;
 110
 111        /* I am lazy... (-8 */
 112#define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value)       \
 113        if (!strcmp(p, option)) {                                       \
 114                if (!v || !*v)                                          \
 115                        variable = default_value;                       \
 116                else {                                                  \
 117                        variable = simple_strtoul(ov = v, &v, 0);       \
 118                        if (*v)                                         \
 119                                goto needs_val;                         \
 120                }                                                       \
 121        }
 122#define NTFS_GETOPT(option, variable)                                   \
 123        if (!strcmp(p, option)) {                                       \
 124                if (!v || !*v)                                          \
 125                        goto needs_arg;                                 \
 126                variable = simple_strtoul(ov = v, &v, 0);               \
 127                if (*v)                                                 \
 128                        goto needs_val;                                 \
 129        }
 130#define NTFS_GETOPT_OCTAL(option, variable)                             \
 131        if (!strcmp(p, option)) {                                       \
 132                if (!v || !*v)                                          \
 133                        goto needs_arg;                                 \
 134                variable = simple_strtoul(ov = v, &v, 8);               \
 135                if (*v)                                                 \
 136                        goto needs_val;                                 \
 137        }
 138#define NTFS_GETOPT_BOOL(option, variable)                              \
 139        if (!strcmp(p, option)) {                                       \
 140                bool val;                                               \
 141                if (!simple_getbool(v, &val))                           \
 142                        goto needs_bool;                                \
 143                variable = val;                                         \
 144        }
 145#define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array)          \
 146        if (!strcmp(p, option)) {                                       \
 147                int _i;                                                 \
 148                if (!v || !*v)                                          \
 149                        goto needs_arg;                                 \
 150                ov = v;                                                 \
 151                if (variable == -1)                                     \
 152                        variable = 0;                                   \
 153                for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
 154                        if (!strcmp(opt_array[_i].str, v)) {            \
 155                                variable |= opt_array[_i].val;          \
 156                                break;                                  \
 157                        }                                               \
 158                if (!opt_array[_i].str || !*opt_array[_i].str)          \
 159                        goto needs_val;                                 \
 160        }
 161        if (!opt || !*opt)
 162                goto no_mount_options;
 163        ntfs_debug("Entering with mount options string: %s", opt);
 164        while ((p = strsep(&opt, ","))) {
 165                if ((v = strchr(p, '=')))
 166                        *v++ = 0;
 167                NTFS_GETOPT("uid", uid)
 168                else NTFS_GETOPT("gid", gid)
 169                else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
 170                else NTFS_GETOPT_OCTAL("fmask", fmask)
 171                else NTFS_GETOPT_OCTAL("dmask", dmask)
 172                else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
 173                else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
 174                else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
 175                else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
 176                else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
 177                else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
 178                                on_errors_arr)
 179                else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
 180                        ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
 181                                        p);
 182                else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
 183                        if (!strcmp(p, "iocharset"))
 184                                ntfs_warning(vol->sb, "Option iocharset is "
 185                                                "deprecated. Please use "
 186                                                "option nls=<charsetname> in "
 187                                                "the future.");
 188                        if (!v || !*v)
 189                                goto needs_arg;
 190use_utf8:
 191                        old_nls = nls_map;
 192                        nls_map = load_nls(v);
 193                        if (!nls_map) {
 194                                if (!old_nls) {
 195                                        ntfs_error(vol->sb, "NLS character set "
 196                                                        "%s not found.", v);
 197                                        return false;
 198                                }
 199                                ntfs_error(vol->sb, "NLS character set %s not "
 200                                                "found. Using previous one %s.",
 201                                                v, old_nls->charset);
 202                                nls_map = old_nls;
 203                        } else /* nls_map */ {
 204                                unload_nls(old_nls);
 205                        }
 206                } else if (!strcmp(p, "utf8")) {
 207                        bool val = false;
 208                        ntfs_warning(vol->sb, "Option utf8 is no longer "
 209                                   "supported, using option nls=utf8. Please "
 210                                   "use option nls=utf8 in the future and "
 211                                   "make sure utf8 is compiled either as a "
 212                                   "module or into the kernel.");
 213                        if (!v || !*v)
 214                                val = true;
 215                        else if (!simple_getbool(v, &val))
 216                                goto needs_bool;
 217                        if (val) {
 218                                v = utf8;
 219                                goto use_utf8;
 220                        }
 221                } else {
 222                        ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
 223                        if (errors < INT_MAX)
 224                                errors++;
 225                }
 226#undef NTFS_GETOPT_OPTIONS_ARRAY
 227#undef NTFS_GETOPT_BOOL
 228#undef NTFS_GETOPT
 229#undef NTFS_GETOPT_WITH_DEFAULT
 230        }
 231no_mount_options:
 232        if (errors && !sloppy)
 233                return false;
 234        if (sloppy)
 235                ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
 236                                "unrecognized mount option(s) and continuing.");
 237        /* Keep this first! */
 238        if (on_errors != -1) {
 239                if (!on_errors) {
 240                        ntfs_error(vol->sb, "Invalid errors option argument "
 241                                        "or bug in options parser.");
 242                        return false;
 243                }
 244        }
 245        if (nls_map) {
 246                if (vol->nls_map && vol->nls_map != nls_map) {
 247                        ntfs_error(vol->sb, "Cannot change NLS character set "
 248                                        "on remount.");
 249                        return false;
 250                } /* else (!vol->nls_map) */
 251                ntfs_debug("Using NLS character set %s.", nls_map->charset);
 252                vol->nls_map = nls_map;
 253        } else /* (!nls_map) */ {
 254                if (!vol->nls_map) {
 255                        vol->nls_map = load_nls_default();
 256                        if (!vol->nls_map) {
 257                                ntfs_error(vol->sb, "Failed to load default "
 258                                                "NLS character set.");
 259                                return false;
 260                        }
 261                        ntfs_debug("Using default NLS character set (%s).",
 262                                        vol->nls_map->charset);
 263                }
 264        }
 265        if (mft_zone_multiplier != -1) {
 266                if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
 267                                mft_zone_multiplier) {
 268                        ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
 269                                        "on remount.");
 270                        return false;
 271                }
 272                if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
 273                        ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
 274                                        "Using default value, i.e. 1.");
 275                        mft_zone_multiplier = 1;
 276                }
 277                vol->mft_zone_multiplier = mft_zone_multiplier;
 278        }
 279        if (!vol->mft_zone_multiplier)
 280                vol->mft_zone_multiplier = 1;
 281        if (on_errors != -1)
 282                vol->on_errors = on_errors;
 283        if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
 284                vol->on_errors |= ON_ERRORS_CONTINUE;
 285        if (uid != (uid_t)-1)
 286                vol->uid = uid;
 287        if (gid != (gid_t)-1)
 288                vol->gid = gid;
 289        if (fmask != (mode_t)-1)
 290                vol->fmask = fmask;
 291        if (dmask != (mode_t)-1)
 292                vol->dmask = dmask;
 293        if (show_sys_files != -1) {
 294                if (show_sys_files)
 295                        NVolSetShowSystemFiles(vol);
 296                else
 297                        NVolClearShowSystemFiles(vol);
 298        }
 299        if (case_sensitive != -1) {
 300                if (case_sensitive)
 301                        NVolSetCaseSensitive(vol);
 302                else
 303                        NVolClearCaseSensitive(vol);
 304        }
 305        if (disable_sparse != -1) {
 306                if (disable_sparse)
 307                        NVolClearSparseEnabled(vol);
 308                else {
 309                        if (!NVolSparseEnabled(vol) &&
 310                                        vol->major_ver && vol->major_ver < 3)
 311                                ntfs_warning(vol->sb, "Not enabling sparse "
 312                                                "support due to NTFS volume "
 313                                                "version %i.%i (need at least "
 314                                                "version 3.0).", vol->major_ver,
 315                                                vol->minor_ver);
 316                        else
 317                                NVolSetSparseEnabled(vol);
 318                }
 319        }
 320        return true;
 321needs_arg:
 322        ntfs_error(vol->sb, "The %s option requires an argument.", p);
 323        return false;
 324needs_bool:
 325        ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
 326        return false;
 327needs_val:
 328        ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
 329        return false;
 330}
 331
 332#ifdef NTFS_RW
 333
 334/**
 335 * ntfs_write_volume_flags - write new flags to the volume information flags
 336 * @vol:        ntfs volume on which to modify the flags
 337 * @flags:      new flags value for the volume information flags
 338 *
 339 * Internal function.  You probably want to use ntfs_{set,clear}_volume_flags()
 340 * instead (see below).
 341 *
 342 * Replace the volume information flags on the volume @vol with the value
 343 * supplied in @flags.  Note, this overwrites the volume information flags, so
 344 * make sure to combine the flags you want to modify with the old flags and use
 345 * the result when calling ntfs_write_volume_flags().
 346 *
 347 * Return 0 on success and -errno on error.
 348 */
 349static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
 350{
 351        ntfs_inode *ni = NTFS_I(vol->vol_ino);
 352        MFT_RECORD *m;
 353        VOLUME_INFORMATION *vi;
 354        ntfs_attr_search_ctx *ctx;
 355        int err;
 356
 357        ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
 358                        le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
 359        if (vol->vol_flags == flags)
 360                goto done;
 361        BUG_ON(!ni);
 362        m = map_mft_record(ni);
 363        if (IS_ERR(m)) {
 364                err = PTR_ERR(m);
 365                goto err_out;
 366        }
 367        ctx = ntfs_attr_get_search_ctx(ni, m);
 368        if (!ctx) {
 369                err = -ENOMEM;
 370                goto put_unm_err_out;
 371        }
 372        err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
 373                        ctx);
 374        if (err)
 375                goto put_unm_err_out;
 376        vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
 377                        le16_to_cpu(ctx->attr->data.resident.value_offset));
 378        vol->vol_flags = vi->flags = flags;
 379        flush_dcache_mft_record_page(ctx->ntfs_ino);
 380        mark_mft_record_dirty(ctx->ntfs_ino);
 381        ntfs_attr_put_search_ctx(ctx);
 382        unmap_mft_record(ni);
 383done:
 384        ntfs_debug("Done.");
 385        return 0;
 386put_unm_err_out:
 387        if (ctx)
 388                ntfs_attr_put_search_ctx(ctx);
 389        unmap_mft_record(ni);
 390err_out:
 391        ntfs_error(vol->sb, "Failed with error code %i.", -err);
 392        return err;
 393}
 394
 395/**
 396 * ntfs_set_volume_flags - set bits in the volume information flags
 397 * @vol:        ntfs volume on which to modify the flags
 398 * @flags:      flags to set on the volume
 399 *
 400 * Set the bits in @flags in the volume information flags on the volume @vol.
 401 *
 402 * Return 0 on success and -errno on error.
 403 */
 404static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
 405{
 406        flags &= VOLUME_FLAGS_MASK;
 407        return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
 408}
 409
 410/**
 411 * ntfs_clear_volume_flags - clear bits in the volume information flags
 412 * @vol:        ntfs volume on which to modify the flags
 413 * @flags:      flags to clear on the volume
 414 *
 415 * Clear the bits in @flags in the volume information flags on the volume @vol.
 416 *
 417 * Return 0 on success and -errno on error.
 418 */
 419static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
 420{
 421        flags &= VOLUME_FLAGS_MASK;
 422        flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
 423        return ntfs_write_volume_flags(vol, flags);
 424}
 425
 426#endif /* NTFS_RW */
 427
 428/**
 429 * ntfs_remount - change the mount options of a mounted ntfs filesystem
 430 * @sb:         superblock of mounted ntfs filesystem
 431 * @flags:      remount flags
 432 * @opt:        remount options string
 433 *
 434 * Change the mount options of an already mounted ntfs filesystem.
 435 *
 436 * NOTE:  The VFS sets the @sb->s_flags remount flags to @flags after
 437 * ntfs_remount() returns successfully (i.e. returns 0).  Otherwise,
 438 * @sb->s_flags are not changed.
 439 */
 440static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
 441{
 442        ntfs_volume *vol = NTFS_SB(sb);
 443
 444        ntfs_debug("Entering with remount options string: %s", opt);
 445
 446        lock_kernel();
 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 occured.
 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                        unlock_kernel();
 471                        return -EROFS;
 472                }
 473                if (vol->vol_flags & VOLUME_IS_DIRTY) {
 474                        ntfs_error(sb, "Volume is dirty and read-only%s", es);
 475                        unlock_kernel();
 476                        return -EROFS;
 477                }
 478                if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
 479                        ntfs_error(sb, "Volume has been modified by chkdsk "
 480                                        "and is read-only%s", es);
 481                        unlock_kernel();
 482                        return -EROFS;
 483                }
 484                if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
 485                        ntfs_error(sb, "Volume has unsupported flags set "
 486                                        "(0x%x) and is read-only%s",
 487                                        (unsigned)le16_to_cpu(vol->vol_flags),
 488                                        es);
 489                        unlock_kernel();
 490                        return -EROFS;
 491                }
 492                if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
 493                        ntfs_error(sb, "Failed to set dirty bit in volume "
 494                                        "information flags%s", es);
 495                        unlock_kernel();
 496                        return -EROFS;
 497                }
 498#if 0
 499                // TODO: Enable this code once we start modifying anything that
 500                //       is different between NTFS 1.2 and 3.x...
 501                /* Set NT4 compatibility flag on newer NTFS version volumes. */
 502                if ((vol->major_ver > 1)) {
 503                        if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
 504                                ntfs_error(sb, "Failed to set NT4 "
 505                                                "compatibility flag%s", es);
 506                                NVolSetErrors(vol);
 507                                return -EROFS;
 508                        }
 509                }
 510#endif
 511                if (!ntfs_empty_logfile(vol->logfile_ino)) {
 512                        ntfs_error(sb, "Failed to empty journal $LogFile%s",
 513                                        es);
 514                        NVolSetErrors(vol);
 515                        unlock_kernel();
 516                        return -EROFS;
 517                }
 518                if (!ntfs_mark_quotas_out_of_date(vol)) {
 519                        ntfs_error(sb, "Failed to mark quotas out of date%s",
 520                                        es);
 521                        NVolSetErrors(vol);
 522                        unlock_kernel();
 523                        return -EROFS;
 524                }
 525                if (!ntfs_stamp_usnjrnl(vol)) {
 526                        ntfs_error(sb, "Failed to stamp transation log "
 527                                        "($UsnJrnl)%s", es);
 528                        NVolSetErrors(vol);
 529                        unlock_kernel();
 530                        return -EROFS;
 531                }
 532        } else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
 533                /* Remounting read-only. */
 534                if (!NVolErrors(vol)) {
 535                        if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
 536                                ntfs_warning(sb, "Failed to clear dirty bit "
 537                                                "in volume information "
 538                                                "flags.  Run chkdsk.");
 539                }
 540        }
 541#endif /* NTFS_RW */
 542
 543        // TODO: Deal with *flags.
 544
 545        if (!parse_options(vol, opt)) {
 546                unlock_kernel();
 547                return -EINVAL;
 548        }
 549        unlock_kernel();
 550        ntfs_debug("Done.");
 551        return 0;
 552}
 553
 554/**
 555 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
 556 * @sb:         Super block of the device to which @b belongs.
 557 * @b:          Boot sector of device @sb to check.
 558 * @silent:     If 'true', all output will be silenced.
 559 *
 560 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
 561 * sector. Returns 'true' if it is valid and 'false' if not.
 562 *
 563 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
 564 * is 'true'.
 565 */
 566static bool is_boot_sector_ntfs(const struct super_block *sb,
 567                const NTFS_BOOT_SECTOR *b, const bool silent)
 568{
 569        /*
 570         * Check that checksum == sum of u32 values from b to the checksum
 571         * field.  If checksum is zero, no checking is done.  We will work when
 572         * the checksum test fails, since some utilities update the boot sector
 573         * ignoring the checksum which leaves the checksum out-of-date.  We
 574         * report a warning if this is the case.
 575         */
 576        if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
 577                le32 *u;
 578                u32 i;
 579
 580                for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
 581                        i += le32_to_cpup(u);
 582                if (le32_to_cpu(b->checksum) != i)
 583                        ntfs_warning(sb, "Invalid boot sector checksum.");
 584        }
 585        /* Check OEMidentifier is "NTFS    " */
 586        if (b->oem_id != magicNTFS)
 587                goto not_ntfs;
 588        /* Check bytes per sector value is between 256 and 4096. */
 589        if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
 590                        le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
 591                goto not_ntfs;
 592        /* Check sectors per cluster value is valid. */
 593        switch (b->bpb.sectors_per_cluster) {
 594        case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
 595                break;
 596        default:
 597                goto not_ntfs;
 598        }
 599        /* Check the cluster size is not above the maximum (64kiB). */
 600        if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
 601                        b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
 602                goto not_ntfs;
 603        /* Check reserved/unused fields are really zero. */
 604        if (le16_to_cpu(b->bpb.reserved_sectors) ||
 605                        le16_to_cpu(b->bpb.root_entries) ||
 606                        le16_to_cpu(b->bpb.sectors) ||
 607                        le16_to_cpu(b->bpb.sectors_per_fat) ||
 608                        le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
 609                goto not_ntfs;
 610        /* Check clusters per file mft record value is valid. */
 611        if ((u8)b->clusters_per_mft_record < 0xe1 ||
 612                        (u8)b->clusters_per_mft_record > 0xf7)
 613                switch (b->clusters_per_mft_record) {
 614                case 1: case 2: case 4: case 8: case 16: case 32: case 64:
 615                        break;
 616                default:
 617                        goto not_ntfs;
 618                }
 619        /* Check clusters per index block value is valid. */
 620        if ((u8)b->clusters_per_index_record < 0xe1 ||
 621                        (u8)b->clusters_per_index_record > 0xf7)
 622                switch (b->clusters_per_index_record) {
 623                case 1: case 2: case 4: case 8: case 16: case 32: case 64:
 624                        break;
 625                default:
 626                        goto not_ntfs;
 627                }
 628        /*
 629         * Check for valid end of sector marker. We will work without it, but
 630         * many BIOSes will refuse to boot from a bootsector if the magic is
 631         * incorrect, so we emit a warning.
 632         */
 633        if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
 634                ntfs_warning(sb, "Invalid end of sector marker.");
 635        return true;
 636not_ntfs:
 637        return false;
 638}
 639
 640/**
 641 * read_ntfs_boot_sector - read the NTFS boot sector of a device
 642 * @sb:         super block of device to read the boot sector from
 643 * @silent:     if true, suppress all output
 644 *
 645 * Reads the boot sector from the device and validates it. If that fails, tries
 646 * to read the backup boot sector, first from the end of the device a-la NT4 and
 647 * later and then from the middle of the device a-la NT3.51 and before.
 648 *
 649 * If a valid boot sector is found but it is not the primary boot sector, we
 650 * repair the primary boot sector silently (unless the device is read-only or
 651 * the primary boot sector is not accessible).
 652 *
 653 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
 654 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
 655 * to their respective values.
 656 *
 657 * Return the unlocked buffer head containing the boot sector or NULL on error.
 658 */
 659static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
 660                const int silent)
 661{
 662        const char *read_err_str = "Unable to read %s boot sector.";
 663        struct buffer_head *bh_primary, *bh_backup;
 664        sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
 665
 666        /* Try to read primary boot sector. */
 667        if ((bh_primary = sb_bread(sb, 0))) {
 668                if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
 669                                bh_primary->b_data, silent))
 670                        return bh_primary;
 671                if (!silent)
 672                        ntfs_error(sb, "Primary boot sector is invalid.");
 673        } else if (!silent)
 674                ntfs_error(sb, read_err_str, "primary");
 675        if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
 676                if (bh_primary)
 677                        brelse(bh_primary);
 678                if (!silent)
 679                        ntfs_error(sb, "Mount option errors=recover not used. "
 680                                        "Aborting without trying to recover.");
 681                return NULL;
 682        }
 683        /* Try to read NT4+ backup boot sector. */
 684        if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
 685                if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
 686                                bh_backup->b_data, silent))
 687                        goto hotfix_primary_boot_sector;
 688                brelse(bh_backup);
 689        } else if (!silent)
 690                ntfs_error(sb, read_err_str, "backup");
 691        /* Try to read NT3.51- backup boot sector. */
 692        if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
 693                if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
 694                                bh_backup->b_data, silent))
 695                        goto hotfix_primary_boot_sector;
 696                if (!silent)
 697                        ntfs_error(sb, "Could not find a valid backup boot "
 698                                        "sector.");
 699                brelse(bh_backup);
 700        } else if (!silent)
 701                ntfs_error(sb, read_err_str, "backup");
 702        /* We failed. Cleanup and return. */
 703        if (bh_primary)
 704                brelse(bh_primary);
 705        return NULL;
 706hotfix_primary_boot_sector:
 707        if (bh_primary) {
 708                /*
 709                 * If we managed to read sector zero and the volume is not
 710                 * read-only, copy the found, valid backup boot sector to the
 711                 * primary boot sector.  Note we only copy the actual boot
 712                 * sector structure, not the actual whole device sector as that
 713                 * may be bigger and would potentially damage the $Boot system
 714                 * file (FIXME: Would be nice to know if the backup boot sector
 715                 * on a large sector device contains the whole boot loader or
 716                 * just the first 512 bytes).
 717                 */
 718                if (!(sb->s_flags & MS_RDONLY)) {
 719                        ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
 720                                        "boot sector from backup copy.");
 721                        memcpy(bh_primary->b_data, bh_backup->b_data,
 722                                        NTFS_BLOCK_SIZE);
 723                        mark_buffer_dirty(bh_primary);
 724                        sync_dirty_buffer(bh_primary);
 725                        if (buffer_uptodate(bh_primary)) {
 726                                brelse(bh_backup);
 727                                return bh_primary;
 728                        }
 729                        ntfs_error(sb, "Hot-fix: Device write error while "
 730                                        "recovering primary boot sector.");
 731                } else {
 732                        ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
 733                                        "sector failed: Read-only mount.");
 734                }
 735                brelse(bh_primary);
 736        }
 737        ntfs_warning(sb, "Using backup boot sector.");
 738        return bh_backup;
 739}
 740
 741/**
 742 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
 743 * @vol:        volume structure to initialise with data from boot sector
 744 * @b:          boot sector to parse
 745 *
 746 * Parse the ntfs boot sector @b and store all imporant information therein in
 747 * the ntfs super block @vol.  Return 'true' on success and 'false' on error.
 748 */
 749static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
 750{
 751        unsigned int sectors_per_cluster_bits, nr_hidden_sects;
 752        int clusters_per_mft_record, clusters_per_index_record;
 753        s64 ll;
 754
 755        vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
 756        vol->sector_size_bits = ffs(vol->sector_size) - 1;
 757        ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
 758                        vol->sector_size);
 759        ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
 760                        vol->sector_size_bits);
 761        if (vol->sector_size < vol->sb->s_blocksize) {
 762                ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
 763                                "device block size (%lu).  This is not "
 764                                "supported.  Sorry.", vol->sector_size,
 765                                vol->sb->s_blocksize);
 766                return false;
 767        }
 768        ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
 769        sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
 770        ntfs_debug("sectors_per_cluster_bits = 0x%x",
 771                        sectors_per_cluster_bits);
 772        nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
 773        ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
 774        vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
 775        vol->cluster_size_mask = vol->cluster_size - 1;
 776        vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
 777        ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
 778                        vol->cluster_size);
 779        ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
 780        ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
 781        if (vol->cluster_size < vol->sector_size) {
 782                ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
 783                                "sector size (%i).  This is not supported.  "
 784                                "Sorry.", vol->cluster_size, vol->sector_size);
 785                return false;
 786        }
 787        clusters_per_mft_record = b->clusters_per_mft_record;
 788        ntfs_debug("clusters_per_mft_record = %i (0x%x)",
 789                        clusters_per_mft_record, clusters_per_mft_record);
 790        if (clusters_per_mft_record > 0)
 791                vol->mft_record_size = vol->cluster_size <<
 792                                (ffs(clusters_per_mft_record) - 1);
 793        else
 794                /*
 795                 * When mft_record_size < cluster_size, clusters_per_mft_record
 796                 * = -log2(mft_record_size) bytes. mft_record_size normaly is
 797                 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
 798                 */
 799                vol->mft_record_size = 1 << -clusters_per_mft_record;
 800        vol->mft_record_size_mask = vol->mft_record_size - 1;
 801        vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
 802        ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
 803                        vol->mft_record_size);
 804        ntfs_debug("vol->mft_record_size_mask = 0x%x",
 805                        vol->mft_record_size_mask);
 806        ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
 807                        vol->mft_record_size_bits, vol->mft_record_size_bits);
 808        /*
 809         * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
 810         * we store $MFT/$DATA, the table of mft records in the page cache.
 811         */
 812        if (vol->mft_record_size > PAGE_CACHE_SIZE) {
 813                ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
 814                                "PAGE_CACHE_SIZE on your system (%lu).  "
 815                                "This is not supported.  Sorry.",
 816                                vol->mft_record_size, PAGE_CACHE_SIZE);
 817                return false;
 818        }
 819        /* We cannot support mft record sizes below the sector size. */
 820        if (vol->mft_record_size < vol->sector_size) {
 821                ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
 822                                "sector size (%i).  This is not supported.  "
 823                                "Sorry.", vol->mft_record_size,
 824                                vol->sector_size);
 825                return false;
 826        }
 827        clusters_per_index_record = b->clusters_per_index_record;
 828        ntfs_debug("clusters_per_index_record = %i (0x%x)",
 829                        clusters_per_index_record, clusters_per_index_record);
 830        if (clusters_per_index_record > 0)
 831                vol->index_record_size = vol->cluster_size <<
 832                                (ffs(clusters_per_index_record) - 1);
 833        else
 834                /*
 835                 * When index_record_size < cluster_size,
 836                 * clusters_per_index_record = -log2(index_record_size) bytes.
 837                 * index_record_size normaly equals 4096 bytes, which is
 838                 * encoded as 0xF4 (-12 in decimal).
 839                 */
 840                vol->index_record_size = 1 << -clusters_per_index_record;
 841        vol->index_record_size_mask = vol->index_record_size - 1;
 842        vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
 843        ntfs_debug("vol->index_record_size = %i (0x%x)",
 844                        vol->index_record_size, vol->index_record_size);
 845        ntfs_debug("vol->index_record_size_mask = 0x%x",
 846                        vol->index_record_size_mask);
 847        ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
 848                        vol->index_record_size_bits,
 849                        vol->index_record_size_bits);
 850        /* We cannot support index record sizes below the sector size. */
 851        if (vol->index_record_size < vol->sector_size) {
 852                ntfs_error(vol->sb, "Index record size (%i) is smaller than "
 853                                "the sector size (%i).  This is not "
 854                                "supported.  Sorry.", vol->index_record_size,
 855                                vol->sector_size);
 856                return false;
 857        }
 858        /*
 859         * Get the size of the volume in clusters and check for 64-bit-ness.
 860         * Windows currently only uses 32 bits to save the clusters so we do
 861         * the same as it is much faster on 32-bit CPUs.
 862         */
 863        ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
 864        if ((u64)ll >= 1ULL << 32) {
 865                ntfs_error(vol->sb, "Cannot handle 64-bit clusters.  Sorry.");
 866                return false;
 867        }
 868        vol->nr_clusters = ll;
 869        ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
 870        /*
 871         * On an architecture where unsigned long is 32-bits, we restrict the
 872         * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
 873         * will hopefully optimize the whole check away.
 874         */
 875        if (sizeof(unsigned long) < 8) {
 876                if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
 877                        ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
 878                                        "large for this architecture.  "
 879                                        "Maximum supported is 2TiB.  Sorry.",
 880                                        (unsigned long long)ll >> (40 -
 881                                        vol->cluster_size_bits));
 882                        return false;
 883                }
 884        }
 885        ll = sle64_to_cpu(b->mft_lcn);
 886        if (ll >= vol->nr_clusters) {
 887                ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
 888                                "volume.  Weird.", (unsigned long long)ll,
 889                                (unsigned long long)ll);
 890                return false;
 891        }
 892        vol->mft_lcn = ll;
 893        ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
 894        ll = sle64_to_cpu(b->mftmirr_lcn);
 895        if (ll >= vol->nr_clusters) {
 896                ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
 897                                "of volume.  Weird.", (unsigned long long)ll,
 898                                (unsigned long long)ll);
 899                return false;
 900        }
 901        vol->mftmirr_lcn = ll;
 902        ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
 903#ifdef NTFS_RW
 904        /*
 905         * Work out the size of the mft mirror in number of mft records. If the
 906         * cluster size is less than or equal to the size taken by four mft
 907         * records, the mft mirror stores the first four mft records. If the
 908         * cluster size is bigger than the size taken by four mft records, the
 909         * mft mirror contains as many mft records as will fit into one
 910         * cluster.
 911         */
 912        if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
 913                vol->mftmirr_size = 4;
 914        else
 915                vol->mftmirr_size = vol->cluster_size >>
 916                                vol->mft_record_size_bits;
 917        ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
 918#endif /* NTFS_RW */
 919        vol->serial_no = le64_to_cpu(b->volume_serial_number);
 920        ntfs_debug("vol->serial_no = 0x%llx",
 921                        (unsigned long long)vol->serial_no);
 922        return true;
 923}
 924
 925/**
 926 * ntfs_setup_allocators - initialize the cluster and mft allocators
 927 * @vol:        volume structure for which to setup the allocators
 928 *
 929 * Setup the cluster (lcn) and mft allocators to the starting values.
 930 */
 931static void ntfs_setup_allocators(ntfs_volume *vol)
 932{
 933#ifdef NTFS_RW
 934        LCN mft_zone_size, mft_lcn;
 935#endif /* NTFS_RW */
 936
 937        ntfs_debug("vol->mft_zone_multiplier = 0x%x",
 938                        vol->mft_zone_multiplier);
 939#ifdef NTFS_RW
 940        /* Determine the size of the MFT zone. */
 941        mft_zone_size = vol->nr_clusters;
 942        switch (vol->mft_zone_multiplier) {  /* % of volume size in clusters */
 943        case 4:
 944                mft_zone_size >>= 1;                    /* 50%   */
 945                break;
 946        case 3:
 947                mft_zone_size = (mft_zone_size +
 948                                (mft_zone_size >> 1)) >> 2;     /* 37.5% */
 949                break;
 950        case 2:
 951                mft_zone_size >>= 2;                    /* 25%   */
 952                break;
 953        /* case 1: */
 954        default:
 955                mft_zone_size >>= 3;                    /* 12.5% */
 956                break;
 957        }
 958        /* Setup the mft zone. */
 959        vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
 960        ntfs_debug("vol->mft_zone_pos = 0x%llx",
 961                        (unsigned long long)vol->mft_zone_pos);
 962        /*
 963         * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
 964         * source) and if the actual mft_lcn is in the expected place or even
 965         * further to the front of the volume, extend the mft_zone to cover the
 966         * beginning of the volume as well.  This is in order to protect the
 967         * area reserved for the mft bitmap as well within the mft_zone itself.
 968         * On non-standard volumes we do not protect it as the overhead would
 969         * be higher than the speed increase we would get by doing it.
 970         */
 971        mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
 972        if (mft_lcn * vol->cluster_size < 16 * 1024)
 973                mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
 974                                vol->cluster_size;
 975        if (vol->mft_zone_start <= mft_lcn)
 976                vol->mft_zone_start = 0;
 977        ntfs_debug("vol->mft_zone_start = 0x%llx",
 978                        (unsigned long long)vol->mft_zone_start);
 979        /*
 980         * Need to cap the mft zone on non-standard volumes so that it does
 981         * not point outside the boundaries of the volume.  We do this by
 982         * halving the zone size until we are inside the volume.
 983         */
 984        vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
 985        while (vol->mft_zone_end >= vol->nr_clusters) {
 986                mft_zone_size >>= 1;
 987                vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
 988        }
 989        ntfs_debug("vol->mft_zone_end = 0x%llx",
 990                        (unsigned long long)vol->mft_zone_end);
 991        /*
 992         * Set the current position within each data zone to the start of the
 993         * respective zone.
 994         */
 995        vol->data1_zone_pos = vol->mft_zone_end;
 996        ntfs_debug("vol->data1_zone_pos = 0x%llx",
 997                        (unsigned long long)vol->data1_zone_pos);
 998        vol->data2_zone_pos = 0;
 999        ntfs_debug("vol->data2_zone_pos = 0x%llx",
1000                        (unsigned long long)vol->data2_zone_pos);
1001
1002        /* Set the mft data allocation position to mft record 24. */
1003        vol->mft_data_pos = 24;
1004        ntfs_debug("vol->mft_data_pos = 0x%llx",
1005                        (unsigned long long)vol->mft_data_pos);
1006#endif /* NTFS_RW */
1007}
1008
1009#ifdef NTFS_RW
1010
1011/**
1012 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
1013 * @vol:        ntfs super block describing device whose mft mirror to load
1014 *
1015 * Return 'true' on success or 'false' on error.
1016 */
1017static bool load_and_init_mft_mirror(ntfs_volume *vol)
1018{
1019        struct inode *tmp_ino;
1020        ntfs_inode *tmp_ni;
1021
1022        ntfs_debug("Entering.");
1023        /* Get mft mirror inode. */
1024        tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
1025        if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1026                if (!IS_ERR(tmp_ino))
1027                        iput(tmp_ino);
1028                /* Caller will display error message. */
1029                return false;
1030        }
1031        /*
1032         * Re-initialize some specifics about $MFTMirr's inode as
1033         * ntfs_read_inode() will have set up the default ones.
1034         */
1035        /* Set uid and gid to root. */
1036        tmp_ino->i_uid = tmp_ino->i_gid = 0;
1037        /* Regular file.  No access for anyone. */
1038        tmp_ino->i_mode = S_IFREG;
1039        /* No VFS initiated operations allowed for $MFTMirr. */
1040        tmp_ino->i_op = &ntfs_empty_inode_ops;
1041        tmp_ino->i_fop = &ntfs_empty_file_ops;
1042        /* Put in our special address space operations. */
1043        tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1044        tmp_ni = NTFS_I(tmp_ino);
1045        /* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1046        NInoSetMstProtected(tmp_ni);
1047        NInoSetSparseDisabled(tmp_ni);
1048        /*
1049         * Set up our little cheat allowing us to reuse the async read io
1050         * completion handler for directories.
1051         */
1052        tmp_ni->itype.index.block_size = vol->mft_record_size;
1053        tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1054        vol->mftmirr_ino = tmp_ino;
1055        ntfs_debug("Done.");
1056        return true;
1057}
1058
1059/**
1060 * check_mft_mirror - compare contents of the mft mirror with the mft
1061 * @vol:        ntfs super block describing device whose mft mirror to check
1062 *
1063 * Return 'true' on success or 'false' on error.
1064 *
1065 * Note, this function also results in the mft mirror runlist being completely
1066 * mapped into memory.  The mft mirror write code requires this and will BUG()
1067 * should it find an unmapped runlist element.
1068 */
1069static bool check_mft_mirror(ntfs_volume *vol)
1070{
1071        struct super_block *sb = vol->sb;
1072        ntfs_inode *mirr_ni;
1073        struct page *mft_page, *mirr_page;
1074        u8 *kmft, *kmirr;
1075        runlist_element *rl, rl2[2];
1076        pgoff_t index;
1077        int mrecs_per_page, i;
1078
1079        ntfs_debug("Entering.");
1080        /* Compare contents of $MFT and $MFTMirr. */
1081        mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1082        BUG_ON(!mrecs_per_page);
1083        BUG_ON(!vol->mftmirr_size);
1084        mft_page = mirr_page = NULL;
1085        kmft = kmirr = NULL;
1086        index = i = 0;
1087        do {
1088                u32 bytes;
1089
1090                /* Switch pages if necessary. */
1091                if (!(i % mrecs_per_page)) {
1092                        if (index) {
1093                                ntfs_unmap_page(mft_page);
1094                                ntfs_unmap_page(mirr_page);
1095                        }
1096                        /* Get the $MFT page. */
1097                        mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1098                                        index);
1099                        if (IS_ERR(mft_page)) {
1100                                ntfs_error(sb, "Failed to read $MFT.");
1101                                return false;
1102                        }
1103                        kmft = page_address(mft_page);
1104                        /* Get the $MFTMirr page. */
1105                        mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1106                                        index);
1107                        if (IS_ERR(mirr_page)) {
1108                                ntfs_error(sb, "Failed to read $MFTMirr.");
1109                                goto mft_unmap_out;
1110                        }
1111                        kmirr = page_address(mirr_page);
1112                        ++index;
1113                }
1114                /* Do not check the record if it is not in use. */
1115                if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1116                        /* Make sure the record is ok. */
1117                        if (ntfs_is_baad_recordp((le32*)kmft)) {
1118                                ntfs_error(sb, "Incomplete multi sector "
1119                                                "transfer detected in mft "
1120                                                "record %i.", i);
1121mm_unmap_out:
1122                                ntfs_unmap_page(mirr_page);
1123mft_unmap_out:
1124                                ntfs_unmap_page(mft_page);
1125                                return false;
1126                        }
1127                }
1128                /* Do not check the mirror record if it is not in use. */
1129                if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1130                        if (ntfs_is_baad_recordp((le32*)kmirr)) {
1131                                ntfs_error(sb, "Incomplete multi sector "
1132                                                "transfer detected in mft "
1133                                                "mirror record %i.", i);
1134                                goto mm_unmap_out;
1135                        }
1136                }
1137                /* Get the amount of data in the current record. */
1138                bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1139                if (bytes < sizeof(MFT_RECORD_OLD) ||
1140                                bytes > vol->mft_record_size ||
1141                                ntfs_is_baad_recordp((le32*)kmft)) {
1142                        bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1143                        if (bytes < sizeof(MFT_RECORD_OLD) ||
1144                                        bytes > vol->mft_record_size ||
1145                                        ntfs_is_baad_recordp((le32*)kmirr))
1146                                bytes = vol->mft_record_size;
1147                }
1148                /* Compare the two records. */
1149                if (memcmp(kmft, kmirr, bytes)) {
1150                        ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1151                                        "match.  Run ntfsfix or chkdsk.", i);
1152                        goto mm_unmap_out;
1153                }
1154                kmft += vol->mft_record_size;
1155                kmirr += vol->mft_record_size;
1156        } while (++i < vol->mftmirr_size);
1157        /* Release the last pages. */
1158        ntfs_unmap_page(mft_page);
1159        ntfs_unmap_page(mirr_page);
1160
1161        /* Construct the mft mirror runlist by hand. */
1162        rl2[0].vcn = 0;
1163        rl2[0].lcn = vol->mftmirr_lcn;
1164        rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1165                        vol->cluster_size - 1) / vol->cluster_size;
1166        rl2[1].vcn = rl2[0].length;
1167        rl2[1].lcn = LCN_ENOENT;
1168        rl2[1].length = 0;
1169        /*
1170         * Because we have just read all of the mft mirror, we know we have
1171         * mapped the full runlist for it.
1172         */
1173        mirr_ni = NTFS_I(vol->mftmirr_ino);
1174        down_read(&mirr_ni->runlist.lock);
1175        rl = mirr_ni->runlist.rl;
1176        /* Compare the two runlists.  They must be identical. */
1177        i = 0;
1178        do {
1179                if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1180                                rl2[i].length != rl[i].length) {
1181                        ntfs_error(sb, "$MFTMirr location mismatch.  "
1182                                        "Run chkdsk.");
1183                        up_read(&mirr_ni->runlist.lock);
1184                        return false;
1185                }
1186        } while (rl2[i++].length);
1187        up_read(&mirr_ni->runlist.lock);
1188        ntfs_debug("Done.");
1189        return true;
1190}
1191
1192/**
1193 * load_and_check_logfile - load and check the logfile inode for a volume
1194 * @vol:        ntfs super block describing device whose logfile to load
1195 *
1196 * Return 'true' on success or 'false' on error.
1197 */
1198static bool load_and_check_logfile(ntfs_volume *vol,
1199                RESTART_PAGE_HEADER **rp)
1200{
1201        struct inode *tmp_ino;
1202
1203        ntfs_debug("Entering.");
1204        tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1205        if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1206                if (!IS_ERR(tmp_ino))
1207                        iput(tmp_ino);
1208                /* Caller will display error message. */
1209                return false;
1210        }
1211        if (!ntfs_check_logfile(tmp_ino, rp)) {
1212                iput(tmp_ino);
1213                /* ntfs_check_logfile() will have displayed error output. */
1214                return false;
1215        }
1216        NInoSetSparseDisabled(NTFS_I(tmp_ino));
1217        vol->logfile_ino = tmp_ino;
1218        ntfs_debug("Done.");
1219        return true;
1220}
1221
1222#define NTFS_HIBERFIL_HEADER_SIZE       4096
1223
1224/**
1225 * check_windows_hibernation_status - check if Windows is suspended on a volume
1226 * @vol:        ntfs super block of device to check
1227 *
1228 * Check if Windows is hibernated on the ntfs volume @vol.  This is done by
1229 * looking for the file hiberfil.sys in the root directory of the volume.  If
1230 * the file is not present Windows is definitely not suspended.
1231 *
1232 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1233 * definitely suspended (this volume is not the system volume).  Caveat:  on a
1234 * system with many volumes it is possible that the < 4kiB check is bogus but
1235 * for now this should do fine.
1236 *
1237 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1238 * hiberfil header (which is the first 4kiB).  If this begins with "hibr",
1239 * Windows is definitely suspended.  If it is completely full of zeroes,
1240 * Windows is definitely not hibernated.  Any other case is treated as if
1241 * Windows is suspended.  This caters for the above mentioned caveat of a
1242 * system with many volumes where no "hibr" magic would be present and there is
1243 * no zero header.
1244 *
1245 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1246 * hibernated on the volume, and -errno on error.
1247 */
1248static int check_windows_hibernation_status(ntfs_volume *vol)
1249{
1250        MFT_REF mref;
1251        struct inode *vi;
1252        ntfs_inode *ni;
1253        struct page *page;
1254        u32 *kaddr, *kend;
1255        ntfs_name *name = NULL;
1256        int ret = 1;
1257        static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
1258                        cpu_to_le16('i'), cpu_to_le16('b'),
1259                        cpu_to_le16('e'), cpu_to_le16('r'),
1260                        cpu_to_le16('f'), cpu_to_le16('i'),
1261                        cpu_to_le16('l'), cpu_to_le16('.'),
1262                        cpu_to_le16('s'), cpu_to_le16('y'),
1263                        cpu_to_le16('s'), 0 };
1264
1265        ntfs_debug("Entering.");
1266        /*
1267         * Find the inode number for the hibernation file by looking up the
1268         * filename hiberfil.sys in the root directory.
1269         */
1270        mutex_lock(&vol->root_ino->i_mutex);
1271        mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1272                        &name);
1273        mutex_unlock(&vol->root_ino->i_mutex);
1274        if (IS_ERR_MREF(mref)) {
1275                ret = MREF_ERR(mref);
1276                /* If the file does not exist, Windows is not hibernated. */
1277                if (ret == -ENOENT) {
1278                        ntfs_debug("hiberfil.sys not present.  Windows is not "
1279                                        "hibernated on the volume.");
1280                        return 0;
1281                }
1282                /* A real error occured. */
1283                ntfs_error(vol->sb, "Failed to find inode number for "
1284                                "hiberfil.sys.");
1285                return ret;
1286        }
1287        /* We do not care for the type of match that was found. */
1288        kfree(name);
1289        /* Get the inode. */
1290        vi = ntfs_iget(vol->sb, MREF(mref));
1291        if (IS_ERR(vi) || is_bad_inode(vi)) {
1292                if (!IS_ERR(vi))
1293                        iput(vi);
1294                ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1295                return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1296        }
1297        if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1298                ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx).  "
1299                                "Windows is hibernated on the volume.  This "
1300                                "is not the system volume.", i_size_read(vi));
1301                goto iput_out;
1302        }
1303        ni = NTFS_I(vi);
1304        page = ntfs_map_page(vi->i_mapping, 0);
1305        if (IS_ERR(page)) {
1306                ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1307                ret = PTR_ERR(page);
1308                goto iput_out;
1309        }
1310        kaddr = (u32*)page_address(page);
1311        if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
1312                ntfs_debug("Magic \"hibr\" found in hiberfil.sys.  Windows is "
1313                                "hibernated on the volume.  This is the "
1314                                "system volume.");
1315                goto unm_iput_out;
1316        }
1317        kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1318        do {
1319                if (unlikely(*kaddr)) {
1320                        ntfs_debug("hiberfil.sys is larger than 4kiB "
1321                                        "(0x%llx), does not contain the "
1322                                        "\"hibr\" magic, and does not have a "
1323                                        "zero header.  Windows is hibernated "
1324                                        "on the volume.  This is not the "
1325                                        "system volume.", i_size_read(vi));
1326                        goto unm_iput_out;
1327                }
1328        } while (++kaddr < kend);
1329        ntfs_debug("hiberfil.sys contains a zero header.  Windows is not "
1330                        "hibernated on the volume.  This is the system "
1331                        "volume.");
1332        ret = 0;
1333unm_iput_out:
1334        ntfs_unmap_page(page);
1335iput_out:
1336        iput(vi);
1337        return ret;
1338}
1339
1340/**
1341 * load_and_init_quota - load and setup the quota file for a volume if present
1342 * @vol:        ntfs super block describing device whose quota file to load
1343 *
1344 * Return 'true' on success or 'false' on error.  If $Quota is not present, we
1345 * leave vol->quota_ino as NULL and return success.
1346 */
1347static bool load_and_init_quota(ntfs_volume *vol)
1348{
1349        MFT_REF mref;
1350        struct inode *tmp_ino;
1351        ntfs_name *name = NULL;
1352        static const ntfschar Quota[7] = { cpu_to_le16('$'),
1353                        cpu_to_le16('Q'), cpu_to_le16('u'),
1354                        cpu_to_le16('o'), cpu_to_le16('t'),
1355                        cpu_to_le16('a'), 0 };
1356        static ntfschar Q[3] = { cpu_to_le16('$'),
1357                        cpu_to_le16('Q'), 0 };
1358
1359        ntfs_debug("Entering.");
1360        /*
1361         * Find the inode number for the quota file by looking up the filename
1362         * $Quota in the extended system files directory $Extend.
1363         */
1364        mutex_lock(&vol->extend_ino->i_mutex);
1365        mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1366                        &name);
1367        mutex_unlock(&vol->extend_ino->i_mutex);
1368        if (IS_ERR_MREF(mref)) {
1369                /*
1370                 * If the file does not exist, quotas are disabled and have
1371                 * never been enabled on this volume, just return success.
1372                 */
1373                if (MREF_ERR(mref) == -ENOENT) {
1374                        ntfs_debug("$Quota not present.  Volume does not have "
1375                                        "quotas enabled.");
1376                        /*
1377                         * No need to try to set quotas out of date if they are
1378                         * not enabled.
1379                         */
1380                        NVolSetQuotaOutOfDate(vol);
1381                        return true;
1382                }
1383                /* A real error occured. */
1384                ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1385                return false;
1386        }
1387        /* We do not care for the type of match that was found. */
1388        kfree(name);
1389        /* Get the inode. */
1390        tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1391        if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1392                if (!IS_ERR(tmp_ino))
1393                        iput(tmp_ino);
1394                ntfs_error(vol->sb, "Failed to load $Quota.");
1395                return false;
1396        }
1397        vol->quota_ino = tmp_ino;
1398        /* Get the $Q index allocation attribute. */
1399        tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1400        if (IS_ERR(tmp_ino)) {
1401                ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1402                return false;
1403        }
1404        vol->quota_q_ino = tmp_ino;
1405        ntfs_debug("Done.");
1406        return true;
1407}
1408
1409/**
1410 * load_and_init_usnjrnl - load and setup the transaction log if present
1411 * @vol:        ntfs super block describing device whose usnjrnl file to load
1412 *
1413 * Return 'true' on success or 'false' on error.
1414 *
1415 * If $UsnJrnl is not present or in the process of being disabled, we set
1416 * NVolUsnJrnlStamped() and return success.
1417 *
1418 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1419 * i.e. transaction logging has only just been enabled or the journal has been
1420 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1421 * and return success.
1422 */
1423static bool load_and_init_usnjrnl(ntfs_volume *vol)
1424{
1425        MFT_REF mref;
1426        struct inode *tmp_ino;
1427        ntfs_inode *tmp_ni;
1428        struct page *page;
1429        ntfs_name *name = NULL;
1430        USN_HEADER *uh;
1431        static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
1432                        cpu_to_le16('U'), cpu_to_le16('s'),
1433                        cpu_to_le16('n'), cpu_to_le16('J'),
1434                        cpu_to_le16('r'), cpu_to_le16('n'),
1435                        cpu_to_le16('l'), 0 };
1436        static ntfschar Max[5] = { cpu_to_le16('$'),
1437                        cpu_to_le16('M'), cpu_to_le16('a'),
1438                        cpu_to_le16('x'), 0 };
1439        static ntfschar J[3] = { cpu_to_le16('$'),
1440                        cpu_to_le16('J'), 0 };
1441
1442        ntfs_debug("Entering.");
1443        /*
1444         * Find the inode number for the transaction log file by looking up the
1445         * filename $UsnJrnl in the extended system files directory $Extend.
1446         */
1447        mutex_lock(&vol->extend_ino->i_mutex);
1448        mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1449                        &name);
1450        mutex_unlock(&vol->extend_ino->i_mutex);
1451        if (IS_ERR_MREF(mref)) {
1452                /*
1453                 * If the file does not exist, transaction logging is disabled,
1454                 * just return success.
1455                 */
1456                if (MREF_ERR(mref) == -ENOENT) {
1457                        ntfs_debug("$UsnJrnl not present.  Volume does not "
1458                                        "have transaction logging enabled.");
1459not_enabled:
1460                        /*
1461                         * No need to try to stamp the transaction log if
1462                         * transaction logging is not enabled.
1463                         */
1464                        NVolSetUsnJrnlStamped(vol);
1465                        return true;
1466                }
1467                /* A real error occured. */
1468                ntfs_error(vol->sb, "Failed to find inode number for "
1469                                "$UsnJrnl.");
1470                return false;
1471        }
1472        /* We do not care for the type of match that was found. */
1473        kfree(name);
1474        /* Get the inode. */
1475        tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1476        if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1477                if (!IS_ERR(tmp_ino))
1478                        iput(tmp_ino);
1479                ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1480                return false;
1481        }
1482        vol->usnjrnl_ino = tmp_ino;
1483        /*
1484         * If the transaction log is in the process of being deleted, we can
1485         * ignore it.
1486         */
1487        if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1488                ntfs_debug("$UsnJrnl in the process of being disabled.  "
1489                                "Volume does not have transaction logging "
1490                                "enabled.");
1491                goto not_enabled;
1492        }
1493        /* Get the $DATA/$Max attribute. */
1494        tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1495        if (IS_ERR(tmp_ino)) {
1496                ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1497                                "attribute.");
1498                return false;
1499        }
1500        vol->usnjrnl_max_ino = tmp_ino;
1501        if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1502                ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1503                                "attribute (size is 0x%llx but should be at "
1504                                "least 0x%zx bytes).", i_size_read(tmp_ino),
1505                                sizeof(USN_HEADER));
1506                return false;
1507        }
1508        /* Get the $DATA/$J attribute. */
1509        tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1510        if (IS_ERR(tmp_ino)) {
1511                ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1512                                "attribute.");
1513                return false;
1514        }
1515        vol->usnjrnl_j_ino = tmp_ino;
1516        /* Verify $J is non-resident and sparse. */
1517        tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1518        if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1519                ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1520                                "and/or not sparse.");
1521                return false;
1522        }
1523        /* Read the USN_HEADER from $DATA/$Max. */
1524        page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1525        if (IS_ERR(page)) {
1526                ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1527                                "attribute.");
1528                return false;
1529        }
1530        uh = (USN_HEADER*)page_address(page);
1531        /* Sanity check the $Max. */
1532        if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1533                        sle64_to_cpu(uh->maximum_size))) {
1534                ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1535                                "maximum size (0x%llx).  $UsnJrnl is corrupt.",
1536                                (long long)sle64_to_cpu(uh->allocation_delta),
1537                                (long long)sle64_to_cpu(uh->maximum_size));
1538                ntfs_unmap_page(page);
1539                return false;
1540        }
1541        /*
1542         * If the transaction log has been stamped and nothing has been written
1543         * to it since, we do not need to stamp it.
1544         */
1545        if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1546                        i_size_read(vol->usnjrnl_j_ino))) {
1547                if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1548                                i_size_read(vol->usnjrnl_j_ino))) {
1549                        ntfs_unmap_page(page);
1550                        ntfs_debug("$UsnJrnl is enabled but nothing has been "
1551                                        "logged since it was last stamped.  "
1552                                        "Treating this as if the volume does "
1553                                        "not have transaction logging "
1554                                        "enabled.");
1555                        goto not_enabled;
1556                }
1557                ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1558                                "which is out of bounds (0x%llx).  $UsnJrnl "
1559                                "is corrupt.",
1560                                (long long)sle64_to_cpu(uh->lowest_valid_usn),
1561                                i_size_read(vol->usnjrnl_j_ino));
1562                ntfs_unmap_page(page);
1563                return false;
1564        }
1565        ntfs_unmap_page(page);
1566        ntfs_debug("Done.");
1567        return true;
1568}
1569
1570/**
1571 * load_and_init_attrdef - load the attribute definitions table for a volume
1572 * @vol:        ntfs super block describing device whose attrdef to load
1573 *
1574 * Return 'true' on success or 'false' on error.
1575 */
1576static bool load_and_init_attrdef(ntfs_volume *vol)
1577{
1578        loff_t i_size;
1579        struct super_block *sb = vol->sb;
1580        struct inode *ino;
1581        struct page *page;
1582        pgoff_t index, max_index;
1583        unsigned int size;
1584
1585        ntfs_debug("Entering.");
1586        /* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1587        ino = ntfs_iget(sb, FILE_AttrDef);
1588        if (IS_ERR(ino) || is_bad_inode(ino)) {
1589                if (!IS_ERR(ino))
1590                        iput(ino);
1591                goto failed;
1592        }
1593        NInoSetSparseDisabled(NTFS_I(ino));
1594        /* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1595        i_size = i_size_read(ino);
1596        if (i_size <= 0 || i_size > 0x7fffffff)
1597                goto iput_failed;
1598        vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1599        if (!vol->attrdef)
1600                goto iput_failed;
1601        index = 0;
1602        max_index = i_size >> PAGE_CACHE_SHIFT;
1603        size = PAGE_CACHE_SIZE;
1604        while (index < max_index) {
1605                /* Read the attrdef table and copy it into the linear buffer. */
1606read_partial_attrdef_page:
1607                page = ntfs_map_page(ino->i_mapping, index);
1608                if (IS_ERR(page))
1609                        goto free_iput_failed;
1610                memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1611                                page_address(page), size);
1612                ntfs_unmap_page(page);
1613        };
1614        if (size == PAGE_CACHE_SIZE) {
1615                size = i_size & ~PAGE_CACHE_MASK;
1616                if (size)
1617                        goto read_partial_attrdef_page;
1618        }
1619        vol->attrdef_size = i_size;
1620        ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1621        iput(ino);
1622        return true;
1623free_iput_failed:
1624        ntfs_free(vol->attrdef);
1625        vol->attrdef = NULL;
1626iput_failed:
1627        iput(ino);
1628failed:
1629        ntfs_error(sb, "Failed to initialize attribute definition table.");
1630        return false;
1631}
1632
1633#endif /* NTFS_RW */
1634
1635/**
1636 * load_and_init_upcase - load the upcase table for an ntfs volume
1637 * @vol:        ntfs super block describing device whose upcase to load
1638 *
1639 * Return 'true' on success or 'false' on error.
1640 */
1641static bool load_and_init_upcase(ntfs_volume *vol)
1642{
1643        loff_t i_size;
1644        struct super_block *sb = vol->sb;
1645        struct inode *ino;
1646        struct page *page;
1647        pgoff_t index, max_index;
1648        unsigned int size;
1649        int i, max;
1650
1651        ntfs_debug("Entering.");
1652        /* Read upcase table and setup vol->upcase and vol->upcase_len. */
1653        ino = ntfs_iget(sb, FILE_UpCase);
1654        if (IS_ERR(ino) || is_bad_inode(ino)) {
1655                if (!IS_ERR(ino))
1656                        iput(ino);
1657                goto upcase_failed;
1658        }
1659        /*
1660         * The upcase size must not be above 64k Unicode characters, must not
1661         * be zero and must be a multiple of sizeof(ntfschar).
1662         */
1663        i_size = i_size_read(ino);
1664        if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1665                        i_size > 64ULL * 1024 * sizeof(ntfschar))
1666                goto iput_upcase_failed;
1667        vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1668        if (!vol->upcase)
1669                goto iput_upcase_failed;
1670        index = 0;
1671        max_index = i_size >> PAGE_CACHE_SHIFT;
1672        size = PAGE_CACHE_SIZE;
1673        while (index < max_index) {
1674                /* Read the upcase table and copy it into the linear buffer. */
1675read_partial_upcase_page:
1676                page = ntfs_map_page(ino->i_mapping, index);
1677                if (IS_ERR(page))
1678                        goto iput_upcase_failed;
1679                memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1680                                page_address(page), size);
1681                ntfs_unmap_page(page);
1682        };
1683        if (size == PAGE_CACHE_SIZE) {
1684                size = i_size & ~PAGE_CACHE_MASK;
1685                if (size)
1686                        goto read_partial_upcase_page;
1687        }
1688        vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1689        ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1690                        i_size, 64 * 1024 * sizeof(ntfschar));
1691        iput(ino);
1692        mutex_lock(&ntfs_lock);
1693        if (!default_upcase) {
1694                ntfs_debug("Using volume specified $UpCase since default is "
1695                                "not present.");
1696                mutex_unlock(&ntfs_lock);
1697                return true;
1698        }
1699        max = default_upcase_len;
1700        if (max > vol->upcase_len)
1701                max = vol->upcase_len;
1702        for (i = 0; i < max; i++)
1703                if (vol->upcase[i] != default_upcase[i])
1704                        break;
1705        if (i == max) {
1706                ntfs_free(vol->upcase);
1707                vol->upcase = default_upcase;
1708                vol->upcase_len = max;
1709                ntfs_nr_upcase_users++;
1710                mutex_unlock(&ntfs_lock);
1711                ntfs_debug("Volume specified $UpCase matches default. Using "
1712                                "default.");
1713                return true;
1714        }
1715        mutex_unlock(&ntfs_lock);
1716        ntfs_debug("Using volume specified $UpCase since it does not match "
1717                        "the default.");
1718        return true;
1719iput_upcase_failed:
1720        iput(ino);
1721        ntfs_free(vol->upcase);
1722        vol->upcase = NULL;
1723upcase_failed:
1724        mutex_lock(&ntfs_lock);
1725        if (default_upcase) {
1726                vol->upcase = default_upcase;
1727                vol->upcase_len = default_upcase_len;
1728                ntfs_nr_upcase_users++;
1729                mutex_unlock(&ntfs_lock);
1730                ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1731                                "default.");
1732                return true;
1733        }
1734        mutex_unlock(&ntfs_lock);
1735        ntfs_error(sb, "Failed to initialize upcase table.");
1736        return false;
1737}
1738
1739/*
1740 * The lcn and mft bitmap inodes are NTFS-internal inodes with
1741 * their own special locking rules:
1742 */
1743static struct lock_class_key
1744        lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1745        mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1746
1747/**
1748 * load_system_files - open the system files using normal functions
1749 * @vol:        ntfs super block describing device whose system files to load
1750 *
1751 * Open the system files with normal access functions and complete setting up
1752 * the ntfs super block @vol.
1753 *
1754 * Return 'true' on success or 'false' on error.
1755 */
1756static bool load_system_files(ntfs_volume *vol)
1757{
1758        struct super_block *sb = vol->sb;
1759        MFT_RECORD *m;
1760        VOLUME_INFORMATION *vi;
1761        ntfs_attr_search_ctx *ctx;
1762#ifdef NTFS_RW
1763        RESTART_PAGE_HEADER *rp;
1764        int err;
1765#endif /* NTFS_RW */
1766
1767        ntfs_debug("Entering.");
1768#ifdef NTFS_RW
1769        /* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1770        if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1771                static const char *es1 = "Failed to load $MFTMirr";
1772                static const char *es2 = "$MFTMirr does not match $MFT";
1773                static const char *es3 = ".  Run ntfsfix and/or chkdsk.";
1774
1775                /* If a read-write mount, convert it to a read-only mount. */
1776                if (!(sb->s_flags & MS_RDONLY)) {
1777                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1778                                        ON_ERRORS_CONTINUE))) {
1779                                ntfs_error(sb, "%s and neither on_errors="
1780                                                "continue nor on_errors="
1781                                                "remount-ro was specified%s",
1782                                                !vol->mftmirr_ino ? es1 : es2,
1783                                                es3);
1784                                goto iput_mirr_err_out;
1785                        }
1786                        sb->s_flags |= MS_RDONLY;
1787                        ntfs_error(sb, "%s.  Mounting read-only%s",
1788                                        !vol->mftmirr_ino ? es1 : es2, es3);
1789                } else
1790                        ntfs_warning(sb, "%s.  Will not be able to remount "
1791                                        "read-write%s",
1792                                        !vol->mftmirr_ino ? es1 : es2, es3);
1793                /* This will prevent a read-write remount. */
1794                NVolSetErrors(vol);
1795        }
1796#endif /* NTFS_RW */
1797        /* Get mft bitmap attribute inode. */
1798        vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1799        if (IS_ERR(vol->mftbmp_ino)) {
1800                ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1801                goto iput_mirr_err_out;
1802        }
1803        lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1804                           &mftbmp_runlist_lock_key);
1805        lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1806                           &mftbmp_mrec_lock_key);
1807        /* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1808        if (!load_and_init_upcase(vol))
1809                goto iput_mftbmp_err_out;
1810#ifdef NTFS_RW
1811        /*
1812         * Read attribute definitions table and setup @vol->attrdef and
1813         * @vol->attrdef_size.
1814         */
1815        if (!load_and_init_attrdef(vol))
1816                goto iput_upcase_err_out;
1817#endif /* NTFS_RW */
1818        /*
1819         * Get the cluster allocation bitmap inode and verify the size, no
1820         * need for any locking at this stage as we are already running
1821         * exclusively as we are mount in progress task.
1822         */
1823        vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1824        if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1825                if (!IS_ERR(vol->lcnbmp_ino))
1826                        iput(vol->lcnbmp_ino);
1827                goto bitmap_failed;
1828        }
1829        lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1830                           &lcnbmp_runlist_lock_key);
1831        lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1832                           &lcnbmp_mrec_lock_key);
1833
1834        NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1835        if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1836                iput(vol->lcnbmp_ino);
1837bitmap_failed:
1838                ntfs_error(sb, "Failed to load $Bitmap.");
1839                goto iput_attrdef_err_out;
1840        }
1841        /*
1842         * Get the volume inode and setup our cache of the volume flags and
1843         * version.
1844         */
1845        vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1846        if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1847                if (!IS_ERR(vol->vol_ino))
1848                        iput(vol->vol_ino);
1849volume_failed:
1850                ntfs_error(sb, "Failed to load $Volume.");
1851                goto iput_lcnbmp_err_out;
1852        }
1853        m = map_mft_record(NTFS_I(vol->vol_ino));
1854        if (IS_ERR(m)) {
1855iput_volume_failed:
1856                iput(vol->vol_ino);
1857                goto volume_failed;
1858        }
1859        if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1860                ntfs_error(sb, "Failed to get attribute search context.");
1861                goto get_ctx_vol_failed;
1862        }
1863        if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1864                        ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1865err_put_vol:
1866                ntfs_attr_put_search_ctx(ctx);
1867get_ctx_vol_failed:
1868                unmap_mft_record(NTFS_I(vol->vol_ino));
1869                goto iput_volume_failed;
1870        }
1871        vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1872                        le16_to_cpu(ctx->attr->data.resident.value_offset));
1873        /* Some bounds checks. */
1874        if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1875                        le32_to_cpu(ctx->attr->data.resident.value_length) >
1876                        (u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1877                goto err_put_vol;
1878        /* Copy the volume flags and version to the ntfs_volume structure. */
1879        vol->vol_flags = vi->flags;
1880        vol->major_ver = vi->major_ver;
1881        vol->minor_ver = vi->minor_ver;
1882        ntfs_attr_put_search_ctx(ctx);
1883        unmap_mft_record(NTFS_I(vol->vol_ino));
1884        printk(KERN_INFO "NTFS volume version %i.%i.\n", vol->major_ver,
1885                        vol->minor_ver);
1886        if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1887                ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1888                                "volume version %i.%i (need at least version "
1889                                "3.0).", vol->major_ver, vol->minor_ver);
1890                NVolClearSparseEnabled(vol);
1891        }
1892#ifdef NTFS_RW
1893        /* Make sure that no unsupported volume flags are set. */
1894        if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1895                static const char *es1a = "Volume is dirty";
1896                static const char *es1b = "Volume has been modified by chkdsk";
1897                static const char *es1c = "Volume has unsupported flags set";
1898                static const char *es2a = ".  Run chkdsk and mount in Windows.";
1899                static const char *es2b = ".  Mount in Windows.";
1900                const char *es1, *es2;
1901
1902                es2 = es2a;
1903                if (vol->vol_flags & VOLUME_IS_DIRTY)
1904                        es1 = es1a;
1905                else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1906                        es1 = es1b;
1907                        es2 = es2b;
1908                } else {
1909                        es1 = es1c;
1910                        ntfs_warning(sb, "Unsupported volume flags 0x%x "
1911                                        "encountered.",
1912                                        (unsigned)le16_to_cpu(vol->vol_flags));
1913                }
1914                /* If a read-write mount, convert it to a read-only mount. */
1915                if (!(sb->s_flags & MS_RDONLY)) {
1916                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1917                                        ON_ERRORS_CONTINUE))) {
1918                                ntfs_error(sb, "%s and neither on_errors="
1919                                                "continue nor on_errors="
1920                                                "remount-ro was specified%s",
1921                                                es1, es2);
1922                                goto iput_vol_err_out;
1923                        }
1924                        sb->s_flags |= MS_RDONLY;
1925                        ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1926                } else
1927                        ntfs_warning(sb, "%s.  Will not be able to remount "
1928                                        "read-write%s", es1, es2);
1929                /*
1930                 * Do not set NVolErrors() because ntfs_remount() re-checks the
1931                 * flags which we need to do in case any flags have changed.
1932                 */
1933        }
1934        /*
1935         * Get the inode for the logfile, check it and determine if the volume
1936         * was shutdown cleanly.
1937         */
1938        rp = NULL;
1939        if (!load_and_check_logfile(vol, &rp) ||
1940                        !ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1941                static const char *es1a = "Failed to load $LogFile";
1942                static const char *es1b = "$LogFile is not clean";
1943                static const char *es2 = ".  Mount in Windows.";
1944                const char *es1;
1945
1946                es1 = !vol->logfile_ino ? es1a : es1b;
1947                /* If a read-write mount, convert it to a read-only mount. */
1948                if (!(sb->s_flags & MS_RDONLY)) {
1949                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1950                                        ON_ERRORS_CONTINUE))) {
1951                                ntfs_error(sb, "%s and neither on_errors="
1952                                                "continue nor on_errors="
1953                                                "remount-ro was specified%s",
1954                                                es1, es2);
1955                                if (vol->logfile_ino) {
1956                                        BUG_ON(!rp);
1957                                        ntfs_free(rp);
1958                                }
1959                                goto iput_logfile_err_out;
1960                        }
1961                        sb->s_flags |= MS_RDONLY;
1962                        ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1963                } else
1964                        ntfs_warning(sb, "%s.  Will not be able to remount "
1965                                        "read-write%s", es1, es2);
1966                /* This will prevent a read-write remount. */
1967                NVolSetErrors(vol);
1968        }
1969        ntfs_free(rp);
1970#endif /* NTFS_RW */
1971        /* Get the root directory inode so we can do path lookups. */
1972        vol->root_ino = ntfs_iget(sb, FILE_root);
1973        if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1974                if (!IS_ERR(vol->root_ino))
1975                        iput(vol->root_ino);
1976                ntfs_error(sb, "Failed to load root directory.");
1977                goto iput_logfile_err_out;
1978        }
1979#ifdef NTFS_RW
1980        /*
1981         * Check if Windows is suspended to disk on the target volume.  If it
1982         * is hibernated, we must not write *anything* to the disk so set
1983         * NVolErrors() without setting the dirty volume flag and mount
1984         * read-only.  This will prevent read-write remounting and it will also
1985         * prevent all writes.
1986         */
1987        err = check_windows_hibernation_status(vol);
1988        if (unlikely(err)) {
1989                static const char *es1a = "Failed to determine if Windows is "
1990                                "hibernated";
1991                static const char *es1b = "Windows is hibernated";
1992                static const char *es2 = ".  Run chkdsk.";
1993                const char *es1;
1994
1995                es1 = err < 0 ? es1a : es1b;
1996                /* If a read-write mount, convert it to a read-only mount. */
1997                if (!(sb->s_flags & MS_RDONLY)) {
1998                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1999                                        ON_ERRORS_CONTINUE))) {
2000                                ntfs_error(sb, "%s and neither on_errors="
2001                                                "continue nor on_errors="
2002                                                "remount-ro was specified%s",
2003                                                es1, es2);
2004                                goto iput_root_err_out;
2005                        }
2006                        sb->s_flags |= MS_RDONLY;
2007                        ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2008                } else
2009                        ntfs_warning(sb, "%s.  Will not be able to remount "
2010                                        "read-write%s", es1, es2);
2011                /* This will prevent a read-write remount. */
2012                NVolSetErrors(vol);
2013        }
2014        /* If (still) a read-write mount, mark the volume dirty. */
2015        if (!(sb->s_flags & MS_RDONLY) &&
2016                        ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
2017                static const char *es1 = "Failed to set dirty bit in volume "
2018                                "information flags";
2019                static const char *es2 = ".  Run chkdsk.";
2020
2021                /* Convert to a read-only mount. */
2022                if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2023                                ON_ERRORS_CONTINUE))) {
2024                        ntfs_error(sb, "%s and neither on_errors=continue nor "
2025                                        "on_errors=remount-ro was specified%s",
2026                                        es1, es2);
2027                        goto iput_root_err_out;
2028                }
2029                ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2030                sb->s_flags |= MS_RDONLY;
2031                /*
2032                 * Do not set NVolErrors() because ntfs_remount() might manage
2033                 * to set the dirty flag in which case all would be well.
2034                 */
2035        }
2036#if 0
2037        // TODO: Enable this code once we start modifying anything that is
2038        //       different between NTFS 1.2 and 3.x...
2039        /*
2040         * If (still) a read-write mount, set the NT4 compatibility flag on
2041         * newer NTFS version volumes.
2042         */
2043        if (!(sb->s_flags & MS_RDONLY) && (vol->major_ver > 1) &&
2044                        ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
2045                static const char *es1 = "Failed to set NT4 compatibility flag";
2046                static const char *es2 = ".  Run chkdsk.";
2047
2048                /* Convert to a read-only mount. */
2049                if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2050                                ON_ERRORS_CONTINUE))) {
2051                        ntfs_error(sb, "%s and neither on_errors=continue nor "
2052                                        "on_errors=remount-ro was specified%s",
2053                                        es1, es2);
2054                        goto iput_root_err_out;
2055                }
2056                ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2057                sb->s_flags |= MS_RDONLY;
2058                NVolSetErrors(vol);
2059        }
2060#endif
2061        /* If (still) a read-write mount, empty the logfile. */
2062        if (!(sb->s_flags & MS_RDONLY) &&
2063                        !ntfs_empty_logfile(vol->logfile_ino)) {
2064                static const char *es1 = "Failed to empty $LogFile";
2065                static const char *es2 = ".  Mount in Windows.";
2066
2067                /* Convert to a read-only mount. */
2068                if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2069                                ON_ERRORS_CONTINUE))) {
2070                        ntfs_error(sb, "%s and neither on_errors=continue nor "
2071                                        "on_errors=remount-ro was specified%s",
2072                                        es1, es2);
2073                        goto iput_root_err_out;
2074                }
2075                ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2076                sb->s_flags |= MS_RDONLY;
2077                NVolSetErrors(vol);
2078        }
2079#endif /* NTFS_RW */
2080        /* If on NTFS versions before 3.0, we are done. */
2081        if (unlikely(vol->major_ver < 3))
2082                return true;
2083        /* NTFS 3.0+ specific initialization. */
2084        /* Get the security descriptors inode. */
2085        vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2086        if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2087                if (!IS_ERR(vol->secure_ino))
2088                        iput(vol->secure_ino);
2089                ntfs_error(sb, "Failed to load $Secure.");
2090                goto iput_root_err_out;
2091        }
2092        // TODO: Initialize security.
2093        /* Get the extended system files' directory inode. */
2094        vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2095        if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2096                if (!IS_ERR(vol->extend_ino))
2097                        iput(vol->extend_ino);
2098                ntfs_error(sb, "Failed to load $Extend.");
2099                goto iput_sec_err_out;
2100        }
2101#ifdef NTFS_RW
2102        /* Find the quota file, load it if present, and set it up. */
2103        if (!load_and_init_quota(vol)) {
2104                static const char *es1 = "Failed to load $Quota";
2105                static const char *es2 = ".  Run chkdsk.";
2106
2107                /* If a read-write mount, convert it to a read-only mount. */
2108                if (!(sb->s_flags & MS_RDONLY)) {
2109                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2110                                        ON_ERRORS_CONTINUE))) {
2111                                ntfs_error(sb, "%s and neither on_errors="
2112                                                "continue nor on_errors="
2113                                                "remount-ro was specified%s",
2114                                                es1, es2);
2115                                goto iput_quota_err_out;
2116                        }
2117                        sb->s_flags |= MS_RDONLY;
2118                        ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2119                } else
2120                        ntfs_warning(sb, "%s.  Will not be able to remount "
2121                                        "read-write%s", es1, es2);
2122                /* This will prevent a read-write remount. */
2123                NVolSetErrors(vol);
2124        }
2125        /* If (still) a read-write mount, mark the quotas out of date. */
2126        if (!(sb->s_flags & MS_RDONLY) &&
2127                        !ntfs_mark_quotas_out_of_date(vol)) {
2128                static const char *es1 = "Failed to mark quotas out of date";
2129                static const char *es2 = ".  Run chkdsk.";
2130
2131                /* Convert to a read-only mount. */
2132                if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2133                                ON_ERRORS_CONTINUE))) {
2134                        ntfs_error(sb, "%s and neither on_errors=continue nor "
2135                                        "on_errors=remount-ro was specified%s",
2136                                        es1, es2);
2137                        goto iput_quota_err_out;
2138                }
2139                ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2140                sb->s_flags |= MS_RDONLY;
2141                NVolSetErrors(vol);
2142        }
2143        /*
2144         * Find the transaction log file ($UsnJrnl), load it if present, check
2145         * it, and set it up.
2146         */
2147        if (!load_and_init_usnjrnl(vol)) {
2148                static const char *es1 = "Failed to load $UsnJrnl";
2149                static const char *es2 = ".  Run chkdsk.";
2150
2151                /* If a read-write mount, convert it to a read-only mount. */
2152                if (!(sb->s_flags & MS_RDONLY)) {
2153                        if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2154                                        ON_ERRORS_CONTINUE))) {
2155                                ntfs_error(sb, "%s and neither on_errors="
2156                                                "continue nor on_errors="
2157                                                "remount-ro was specified%s",
2158                                                es1, es2);
2159                                goto iput_usnjrnl_err_out;
2160                        }
2161                        sb->s_flags |= MS_RDONLY;
2162                        ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2163                } else
2164                        ntfs_warning(sb, "%s.  Will not be able to remount "
2165                                        "read-write%s", es1, es2);
2166                /* This will prevent a read-write remount. */
2167                NVolSetErrors(vol);
2168        }
2169        /* If (still) a read-write mount, stamp the transaction log. */
2170        if (!(sb->s_flags & MS_RDONLY) && !ntfs_stamp_usnjrnl(vol)) {
2171                static const char *es1 = "Failed to stamp transaction log "
2172                                "($UsnJrnl)";
2173                static const char *es2 = ".  Run chkdsk.";
2174
2175                /* Convert to a read-only mount. */
2176                if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2177                                ON_ERRORS_CONTINUE))) {
2178                        ntfs_error(sb, "%s and neither on_errors=continue nor "
2179                                        "on_errors=remount-ro was specified%s",
2180                                        es1, es2);
2181                        goto iput_usnjrnl_err_out;
2182                }
2183                ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2184                sb->s_flags |= MS_RDONLY;
2185                NVolSetErrors(vol);
2186        }
2187#endif /* NTFS_RW */
2188        return true;
2189#ifdef NTFS_RW
2190iput_usnjrnl_err_out:
2191        if (vol->usnjrnl_j_ino)
2192                iput(vol->usnjrnl_j_ino);
2193        if (vol->usnjrnl_max_ino)
2194                iput(vol->usnjrnl_max_ino);
2195        if (vol->usnjrnl_ino)
2196                iput(vol->usnjrnl_ino);
2197iput_quota_err_out:
2198        if (vol->quota_q_ino)
2199                iput(vol->quota_q_ino);
2200        if (vol->quota_ino)
2201                iput(vol->quota_ino);
2202        iput(vol->extend_ino);
2203#endif /* NTFS_RW */
2204iput_sec_err_out:
2205        iput(vol->secure_ino);
2206iput_root_err_out:
2207        iput(vol->root_ino);
2208iput_logfile_err_out:
2209#ifdef NTFS_RW
2210        if (vol->logfile_ino)
2211                iput(vol->logfile_ino);
2212iput_vol_err_out:
2213#endif /* NTFS_RW */
2214        iput(vol->vol_ino);
2215iput_lcnbmp_err_out:
2216        iput(vol->lcnbmp_ino);
2217iput_attrdef_err_out:
2218        vol->attrdef_size = 0;
2219        if (vol->attrdef) {
2220                ntfs_free(vol->attrdef);
2221                vol->attrdef = NULL;
2222        }
2223#ifdef NTFS_RW
2224iput_upcase_err_out:
2225#endif /* NTFS_RW */
2226        vol->upcase_len = 0;
2227        mutex_lock(&ntfs_lock);
2228        if (vol->upcase == default_upcase) {
2229                ntfs_nr_upcase_users--;
2230                vol->upcase = NULL;
2231        }
2232        mutex_unlock(&ntfs_lock);
2233        if (vol->upcase) {
2234                ntfs_free(vol->upcase);
2235                vol->upcase = NULL;
2236        }
2237iput_mftbmp_err_out:
2238        iput(vol->mftbmp_ino);
2239iput_mirr_err_out:
2240#ifdef NTFS_RW
2241        if (vol->mftmirr_ino)
2242                iput(vol->mftmirr_ino);
2243#endif /* NTFS_RW */
2244        return false;
2245}
2246
2247/**
2248 * ntfs_put_super - called by the vfs to unmount a volume
2249 * @sb:         vfs superblock of volume to unmount
2250 *
2251 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2252 * the volume is being unmounted (umount system call has been invoked) and it
2253 * releases all inodes and memory belonging to the NTFS specific part of the
2254 * super block.
2255 */
2256static void ntfs_put_super(struct super_block *sb)
2257{
2258        ntfs_volume *vol = NTFS_SB(sb);
2259
2260        ntfs_debug("Entering.");
2261
2262        lock_kernel();
2263
2264#ifdef NTFS_RW
2265        /*
2266         * Commit all inodes while they are still open in case some of them
2267         * cause others to be dirtied.
2268         */
2269        ntfs_commit_inode(vol->vol_ino);
2270
2271        /* NTFS 3.0+ specific. */
2272        if (vol->major_ver >= 3) {
2273                if (vol->usnjrnl_j_ino)
2274                        ntfs_commit_inode(vol->usnjrnl_j_ino);
2275                if (vol->usnjrnl_max_ino)
2276                        ntfs_commit_inode(vol->usnjrnl_max_ino);
2277                if (vol->usnjrnl_ino)
2278                        ntfs_commit_inode(vol->usnjrnl_ino);
2279                if (vol->quota_q_ino)
2280                        ntfs_commit_inode(vol->quota_q_ino);
2281                if (vol->quota_ino)
2282                        ntfs_commit_inode(vol->quota_ino);
2283                if (vol->extend_ino)
2284                        ntfs_commit_inode(vol->extend_ino);
2285                if (vol->secure_ino)
2286                        ntfs_commit_inode(vol->secure_ino);
2287        }
2288
2289        ntfs_commit_inode(vol->root_ino);
2290
2291        down_write(&vol->lcnbmp_lock);
2292        ntfs_commit_inode(vol->lcnbmp_ino);
2293        up_write(&vol->lcnbmp_lock);
2294
2295        down_write(&vol->mftbmp_lock);
2296        ntfs_commit_inode(vol->mftbmp_ino);
2297        up_write(&vol->mftbmp_lock);
2298
2299        if (vol->logfile_ino)
2300                ntfs_commit_inode(vol->logfile_ino);
2301
2302        if (vol->mftmirr_ino)
2303                ntfs_commit_inode(vol->mftmirr_ino);
2304        ntfs_commit_inode(vol->mft_ino);
2305
2306        /*
2307         * If a read-write mount and no volume errors have occured, mark the
2308         * volume clean.  Also, re-commit all affected inodes.
2309         */
2310        if (!(sb->s_flags & MS_RDONLY)) {
2311                if (!NVolErrors(vol)) {
2312                        if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2313                                ntfs_warning(sb, "Failed to clear dirty bit "
2314                                                "in volume information "
2315                                                "flags.  Run chkdsk.");
2316                        ntfs_commit_inode(vol->vol_ino);
2317                        ntfs_commit_inode(vol->root_ino);
2318                        if (vol->mftmirr_ino)
2319                                ntfs_commit_inode(vol->mftmirr_ino);
2320                        ntfs_commit_inode(vol->mft_ino);
2321                } else {
2322                        ntfs_warning(sb, "Volume has errors.  Leaving volume "
2323                                        "marked dirty.  Run chkdsk.");
2324                }
2325        }
2326#endif /* NTFS_RW */
2327
2328        iput(vol->vol_ino);
2329        vol->vol_ino = NULL;
2330
2331        /* NTFS 3.0+ specific clean up. */
2332        if (vol->major_ver >= 3) {
2333#ifdef NTFS_RW
2334                if (vol->usnjrnl_j_ino) {
2335                        iput(vol->usnjrnl_j_ino);
2336                        vol->usnjrnl_j_ino = NULL;
2337                }
2338                if (vol->usnjrnl_max_ino) {
2339                        iput(vol->usnjrnl_max_ino);
2340                        vol->usnjrnl_max_ino = NULL;
2341                }
2342                if (vol->usnjrnl_ino) {
2343                        iput(vol->usnjrnl_ino);
2344                        vol->usnjrnl_ino = NULL;
2345                }
2346                if (vol->quota_q_ino) {
2347                        iput(vol->quota_q_ino);
2348                        vol->quota_q_ino = NULL;
2349                }
2350                if (vol->quota_ino) {
2351                        iput(vol->quota_ino);
2352                        vol->quota_ino = NULL;
2353                }
2354#endif /* NTFS_RW */
2355                if (vol->extend_ino) {
2356                        iput(vol->extend_ino);
2357                        vol->extend_ino = NULL;
2358                }
2359                if (vol->secure_ino) {
2360                        iput(vol->secure_ino);
2361                        vol->secure_ino = NULL;
2362                }
2363        }
2364
2365        iput(vol->root_ino);
2366        vol->root_ino = NULL;
2367
2368        down_write(&vol->lcnbmp_lock);
2369        iput(vol->lcnbmp_ino);
2370        vol->lcnbmp_ino = NULL;
2371        up_write(&vol->lcnbmp_lock);
2372
2373        down_write(&vol->mftbmp_lock);
2374        iput(vol->mftbmp_ino);
2375        vol->mftbmp_ino = NULL;
2376        up_write(&vol->mftbmp_lock);
2377
2378#ifdef NTFS_RW
2379        if (vol->logfile_ino) {
2380                iput(vol->logfile_ino);
2381                vol->logfile_ino = NULL;
2382        }
2383        if (vol->mftmirr_ino) {
2384                /* Re-commit the mft mirror and mft just in case. */
2385                ntfs_commit_inode(vol->mftmirr_ino);
2386                ntfs_commit_inode(vol->mft_ino);
2387                iput(vol->mftmirr_ino);
2388                vol->mftmirr_ino = NULL;
2389        }
2390        /*
2391         * We should have no dirty inodes left, due to
2392         * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2393         * the underlying mft records are written out and cleaned.
2394         */
2395        ntfs_commit_inode(vol->mft_ino);
2396        write_inode_now(vol->mft_ino, 1);
2397#endif /* NTFS_RW */
2398
2399        iput(vol->mft_ino);
2400        vol->mft_ino = NULL;
2401
2402        /* Throw away the table of attribute definitions. */
2403        vol->attrdef_size = 0;
2404        if (vol->attrdef) {
2405                ntfs_free(vol->attrdef);
2406                vol->attrdef = NULL;
2407        }
2408        vol->upcase_len = 0;
2409        /*
2410         * Destroy the global default upcase table if necessary.  Also decrease
2411         * the number of upcase users if we are a user.
2412         */
2413        mutex_lock(&ntfs_lock);
2414        if (vol->upcase == default_upcase) {
2415                ntfs_nr_upcase_users--;
2416                vol->upcase = NULL;
2417        }
2418        if (!ntfs_nr_upcase_users && default_upcase) {
2419                ntfs_free(default_upcase);
2420                default_upcase = NULL;
2421        }
2422        if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2423                free_compression_buffers();
2424        mutex_unlock(&ntfs_lock);
2425        if (vol->upcase) {
2426                ntfs_free(vol->upcase);
2427                vol->upcase = NULL;
2428        }
2429
2430        unload_nls(vol->nls_map);
2431
2432        sb->s_fs_info = NULL;
2433        kfree(vol);
2434
2435        unlock_kernel();
2436}
2437
2438/**
2439 * get_nr_free_clusters - return the number of free clusters on a volume
2440 * @vol:        ntfs volume for which to obtain free cluster count
2441 *
2442 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2443 * actually calculate the number of clusters in use instead because this
2444 * allows us to not care about partial pages as these will be just zero filled
2445 * and hence not be counted as allocated clusters.
2446 *
2447 * The only particularity is that clusters beyond the end of the logical ntfs
2448 * volume will be marked as allocated to prevent errors which means we have to
2449 * discount those at the end. This is important as the cluster bitmap always
2450 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2451 * the logical volume and marked in use when they are not as they do not exist.
2452 *
2453 * If any pages cannot be read we assume all clusters in the erroring pages are
2454 * in use. This means we return an underestimate on errors which is better than
2455 * an overestimate.
2456 */
2457static s64 get_nr_free_clusters(ntfs_volume *vol)
2458{
2459        s64 nr_free = vol->nr_clusters;
2460        u32 *kaddr;
2461        struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2462        struct page *page;
2463        pgoff_t index, max_index;
2464
2465        ntfs_debug("Entering.");
2466        /* Serialize accesses to the cluster bitmap. */
2467        down_read(&vol->lcnbmp_lock);
2468        /*
2469         * Convert the number of bits into bytes rounded up, then convert into
2470         * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2471         * full and one partial page max_index = 2.
2472         */
2473        max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2474                        PAGE_CACHE_SHIFT;
2475        /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2476        ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2477                        max_index, PAGE_CACHE_SIZE / 4);
2478        for (index = 0; index < max_index; index++) {
2479                unsigned int i;
2480                /*
2481                 * Read the page from page cache, getting it from backing store
2482                 * if necessary, and increment the use count.
2483                 */
2484                page = read_mapping_page(mapping, index, NULL);
2485                /* Ignore pages which errored synchronously. */
2486                if (IS_ERR(page)) {
2487                        ntfs_debug("read_mapping_page() error. Skipping "
2488                                        "page (index 0x%lx).", index);
2489                        nr_free -= PAGE_CACHE_SIZE * 8;
2490                        continue;
2491                }
2492                kaddr = (u32*)kmap_atomic(page, KM_USER0);
2493                /*
2494                 * For each 4 bytes, subtract the number of set bits. If this
2495                 * is the last page and it is partial we don't really care as
2496                 * it just means we do a little extra work but it won't affect
2497                 * the result as all out of range bytes are set to zero by
2498                 * ntfs_readpage().
2499                 */
2500                for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2501                        nr_free -= (s64)hweight32(kaddr[i]);
2502                kunmap_atomic(kaddr, KM_USER0);
2503                page_cache_release(page);
2504        }
2505        ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2506        /*
2507         * Fixup for eventual bits outside logical ntfs volume (see function
2508         * description above).
2509         */
2510        if (vol->nr_clusters & 63)
2511                nr_free += 64 - (vol->nr_clusters & 63);
2512        up_read(&vol->lcnbmp_lock);
2513        /* If errors occured we may well have gone below zero, fix this. */
2514        if (nr_free < 0)
2515                nr_free = 0;
2516        ntfs_debug("Exiting.");
2517        return nr_free;
2518}
2519
2520/**
2521 * __get_nr_free_mft_records - return the number of free inodes on a volume
2522 * @vol:        ntfs volume for which to obtain free inode count
2523 * @nr_free:    number of mft records in filesystem
2524 * @max_index:  maximum number of pages containing set bits
2525 *
2526 * Calculate the number of free mft records (inodes) on the mounted NTFS
2527 * volume @vol. We actually calculate the number of mft records in use instead
2528 * because this allows us to not care about partial pages as these will be just
2529 * zero filled and hence not be counted as allocated mft record.
2530 *
2531 * If any pages cannot be read we assume all mft records in the erroring pages
2532 * are in use. This means we return an underestimate on errors which is better
2533 * than an overestimate.
2534 *
2535 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2536 */
2537static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2538                s64 nr_free, const pgoff_t max_index)
2539{
2540        u32 *kaddr;
2541        struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2542        struct page *page;
2543        pgoff_t index;
2544
2545        ntfs_debug("Entering.");
2546        /* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2547        ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2548                        "0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2549        for (index = 0; index < max_index; index++) {
2550                unsigned int i;
2551                /*
2552                 * Read the page from page cache, getting it from backing store
2553                 * if necessary, and increment the use count.
2554                 */
2555                page = read_mapping_page(mapping, index, NULL);
2556                /* Ignore pages which errored synchronously. */
2557                if (IS_ERR(page)) {
2558                        ntfs_debug("read_mapping_page() error. Skipping "
2559                                        "page (index 0x%lx).", index);
2560                        nr_free -= PAGE_CACHE_SIZE * 8;
2561                        continue;
2562                }
2563                kaddr = (u32*)kmap_atomic(page, KM_USER0);
2564                /*
2565                 * For each 4 bytes, subtract the number of set bits. If this
2566                 * is the last page and it is partial we don't really care as
2567                 * it just means we do a little extra work but it won't affect
2568                 * the result as all out of range bytes are set to zero by
2569                 * ntfs_readpage().
2570                 */
2571                for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2572                        nr_free -= (s64)hweight32(kaddr[i]);
2573                kunmap_atomic(kaddr, KM_USER0);
2574                page_cache_release(page);
2575        }
2576        ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2577                        index - 1);
2578        /* If errors occured we may well have gone below zero, fix this. */
2579        if (nr_free < 0)
2580                nr_free = 0;
2581        ntfs_debug("Exiting.");
2582        return nr_free;
2583}
2584
2585/**
2586 * ntfs_statfs - return information about mounted NTFS volume
2587 * @dentry:     dentry from mounted volume
2588 * @sfs:        statfs structure in which to return the information
2589 *
2590 * Return information about the mounted NTFS volume @dentry in the statfs structure
2591 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2592 * called). We interpret the values to be correct of the moment in time at
2593 * which we are called. Most values are variable otherwise and this isn't just
2594 * the free values but the totals as well. For example we can increase the
2595 * total number of file nodes if we run out and we can keep doing this until
2596 * there is no more space on the volume left at all.
2597 *
2598 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2599 * ustat system calls.
2600 *
2601 * Return 0 on success or -errno on error.
2602 */
2603static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2604{
2605        struct super_block *sb = dentry->d_sb;
2606        s64 size;
2607        ntfs_volume *vol = NTFS_SB(sb);
2608        ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2609        pgoff_t max_index;
2610        unsigned long flags;
2611
2612        ntfs_debug("Entering.");
2613        /* Type of filesystem. */
2614        sfs->f_type   = NTFS_SB_MAGIC;
2615        /* Optimal transfer block size. */
2616        sfs->f_bsize  = PAGE_CACHE_SIZE;
2617        /*
2618         * Total data blocks in filesystem in units of f_bsize and since
2619         * inodes are also stored in data blocs ($MFT is a file) this is just
2620         * the total clusters.
2621         */
2622        sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2623                                PAGE_CACHE_SHIFT;
2624        /* Free data blocks in filesystem in units of f_bsize. */
2625        size          = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2626                                PAGE_CACHE_SHIFT;
2627        if (size < 0LL)
2628                size = 0LL;
2629        /* Free blocks avail to non-superuser, same as above on NTFS. */
2630        sfs->f_bavail = sfs->f_bfree = size;
2631        /* Serialize accesses to the inode bitmap. */
2632        down_read(&vol->mftbmp_lock);
2633        read_lock_irqsave(&mft_ni->size_lock, flags);
2634        size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2635        /*
2636         * Convert the maximum number of set bits into bytes rounded up, then
2637         * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2638         * have one full and one partial page max_index = 2.
2639         */
2640        max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2641                        + 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2642        read_unlock_irqrestore(&mft_ni->size_lock, flags);
2643        /* Number of inodes in filesystem (at this point in time). */
2644        sfs->f_files = size;
2645        /* Free inodes in fs (based on current total count). */
2646        sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2647        up_read(&vol->mftbmp_lock);
2648        /*
2649         * File system id. This is extremely *nix flavour dependent and even
2650         * within Linux itself all fs do their own thing. I interpret this to
2651         * mean a unique id associated with the mounted fs and not the id
2652         * associated with the filesystem driver, the latter is already given
2653         * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2654         * volume serial number splitting it into two 32-bit parts. We enter
2655         * the least significant 32-bits in f_fsid[0] and the most significant
2656         * 32-bits in f_fsid[1].
2657         */
2658        sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2659        sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2660        /* Maximum length of filenames. */
2661        sfs->f_namelen     = NTFS_MAX_NAME_LEN;
2662        return 0;
2663}
2664
2665/**
2666 * The complete super operations.
2667 */
2668static const struct super_operations ntfs_sops = {
2669        .alloc_inode    = ntfs_alloc_big_inode,   /* VFS: Allocate new inode. */
2670        .destroy_inode  = ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2671#ifdef NTFS_RW
2672        //.dirty_inode  = NULL,                 /* VFS: Called from
2673        //                                         __mark_inode_dirty(). */
2674        .write_inode    = ntfs_write_inode,     /* VFS: Write dirty inode to
2675                                                   disk. */
2676        //.drop_inode   = NULL,                 /* VFS: Called just after the
2677        //                                         inode reference count has
2678        //                                         been decreased to zero.
2679        //                                         NOTE: The inode lock is
2680        //                                         held. See fs/inode.c::
2681        //                                         generic_drop_inode(). */
2682        //.delete_inode = NULL,                 /* VFS: Delete inode from disk.
2683        //                                         Called when i_count becomes
2684        //                                         0 and i_nlink is also 0. */
2685        //.write_super  = NULL,                 /* Flush dirty super block to
2686        //                                         disk. */
2687        //.sync_fs      = NULL,                 /* ? */
2688        //.write_super_lockfs   = NULL,         /* ? */
2689        //.unlockfs     = NULL,                 /* ? */
2690#endif /* NTFS_RW */
2691        .put_super      = ntfs_put_super,       /* Syscall: umount. */
2692        .statfs         = ntfs_statfs,          /* Syscall: statfs */
2693        .remount_fs     = ntfs_remount,         /* Syscall: mount -o remount. */
2694        .clear_inode    = ntfs_clear_big_inode, /* VFS: Called when an inode is
2695                                                   removed from memory. */
2696        //.umount_begin = NULL,                 /* Forced umount. */
2697        .show_options   = ntfs_show_options,    /* Show mount options in
2698                                                   proc. */
2699};
2700
2701/**
2702 * ntfs_fill_super - mount an ntfs filesystem
2703 * @sb:         super block of ntfs filesystem to mount
2704 * @opt:        string containing the mount options
2705 * @silent:     silence error output
2706 *
2707 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2708 * with the mount otions in @data with the NTFS filesystem.
2709 *
2710 * If @silent is true, remain silent even if errors are detected. This is used
2711 * during bootup, when the kernel tries to mount the root filesystem with all
2712 * registered filesystems one after the other until one succeeds. This implies
2713 * that all filesystems except the correct one will quite correctly and
2714 * expectedly return an error, but nobody wants to see error messages when in
2715 * fact this is what is supposed to happen.
2716 *
2717 * NOTE: @sb->s_flags contains the mount options flags.
2718 */
2719static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2720{
2721        ntfs_volume *vol;
2722        struct buffer_head *bh;
2723        struct inode *tmp_ino;
2724        int blocksize, result;
2725
2726        /*
2727         * We do a pretty difficult piece of bootstrap by reading the
2728         * MFT (and other metadata) from disk into memory. We'll only
2729         * release this metadata during umount, so the locking patterns
2730         * observed during bootstrap do not count. So turn off the
2731         * observation of locking patterns (strictly for this context
2732         * only) while mounting NTFS. [The validator is still active
2733         * otherwise, even for this context: it will for example record
2734         * lock class registrations.]
2735         */
2736        lockdep_off();
2737        ntfs_debug("Entering.");
2738#ifndef NTFS_RW
2739        sb->s_flags |= MS_RDONLY;
2740#endif /* ! NTFS_RW */
2741        /* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2742        sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2743        vol = NTFS_SB(sb);
2744        if (!vol) {
2745                if (!silent)
2746                        ntfs_error(sb, "Allocation of NTFS volume structure "
2747                                        "failed. Aborting mount...");
2748                lockdep_on();
2749                return -ENOMEM;
2750        }
2751        /* Initialize ntfs_volume structure. */
2752        *vol = (ntfs_volume) {
2753                .sb = sb,
2754                /*
2755                 * Default is group and other don't have any access to files or
2756                 * directories while owner has full access. Further, files by
2757                 * default are not executable but directories are of course
2758                 * browseable.
2759                 */
2760                .fmask = 0177,
2761                .dmask = 0077,
2762        };
2763        init_rwsem(&vol->mftbmp_lock);
2764        init_rwsem(&vol->lcnbmp_lock);
2765
2766        unlock_kernel();
2767
2768        /* By default, enable sparse support. */
2769        NVolSetSparseEnabled(vol);
2770
2771        /* Important to get the mount options dealt with now. */
2772        if (!parse_options(vol, (char*)opt))
2773                goto err_out_now;
2774
2775        /* We support sector sizes up to the PAGE_CACHE_SIZE. */
2776        if (bdev_logical_block_size(sb->s_bdev) > PAGE_CACHE_SIZE) {
2777                if (!silent)
2778                        ntfs_error(sb, "Device has unsupported sector size "
2779                                        "(%i).  The maximum supported sector "
2780                                        "size on this architecture is %lu "
2781                                        "bytes.",
2782                                        bdev_logical_block_size(sb->s_bdev),
2783                                        PAGE_CACHE_SIZE);
2784                goto err_out_now;
2785        }
2786        /*
2787         * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2788         * sector size, whichever is bigger.
2789         */
2790        blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2791        if (blocksize < NTFS_BLOCK_SIZE) {
2792                if (!silent)
2793                        ntfs_error(sb, "Unable to set device block size.");
2794                goto err_out_now;
2795        }
2796        BUG_ON(blocksize != sb->s_blocksize);
2797        ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2798                        blocksize, sb->s_blocksize_bits);
2799        /* Determine the size of the device in units of block_size bytes. */
2800        if (!i_size_read(sb->s_bdev->bd_inode)) {
2801                if (!silent)
2802                        ntfs_error(sb, "Unable to determine device size.");
2803                goto err_out_now;
2804        }
2805        vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2806                        sb->s_blocksize_bits;
2807        /* Read the boot sector and return unlocked buffer head to it. */
2808        if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2809                if (!silent)
2810                        ntfs_error(sb, "Not an NTFS volume.");
2811                goto err_out_now;
2812        }
2813        /*
2814         * Extract the data from the boot sector and setup the ntfs volume
2815         * using it.
2816         */
2817        result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2818        brelse(bh);
2819        if (!result) {
2820                if (!silent)
2821                        ntfs_error(sb, "Unsupported NTFS filesystem.");
2822                goto err_out_now;
2823        }
2824        /*
2825         * If the boot sector indicates a sector size bigger than the current
2826         * device block size, switch the device block size to the sector size.
2827         * TODO: It may be possible to support this case even when the set
2828         * below fails, we would just be breaking up the i/o for each sector
2829         * into multiple blocks for i/o purposes but otherwise it should just
2830         * work.  However it is safer to leave disabled until someone hits this
2831         * error message and then we can get them to try it without the setting
2832         * so we know for sure that it works.
2833         */
2834        if (vol->sector_size > blocksize) {
2835                blocksize = sb_set_blocksize(sb, vol->sector_size);
2836                if (blocksize != vol->sector_size) {
2837                        if (!silent)
2838                                ntfs_error(sb, "Unable to set device block "
2839                                                "size to sector size (%i).",
2840                                                vol->sector_size);
2841                        goto err_out_now;
2842                }
2843                BUG_ON(blocksize != sb->s_blocksize);
2844                vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2845                                sb->s_blocksize_bits;
2846                ntfs_debug("Changed device block size to %i bytes (block size "
2847                                "bits %i) to match volume sector size.",
2848                                blocksize, sb->s_blocksize_bits);
2849        }
2850        /* Initialize the cluster and mft allocators. */
2851        ntfs_setup_allocators(vol);
2852        /* Setup remaining fields in the super block. */
2853        sb->s_magic = NTFS_SB_MAGIC;
2854        /*
2855         * Ntfs allows 63 bits for the file size, i.e. correct would be:
2856         *      sb->s_maxbytes = ~0ULL >> 1;
2857         * But the kernel uses a long as the page cache page index which on
2858         * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2859         * defined to the maximum the page cache page index can cope with
2860         * without overflowing the index or to 2^63 - 1, whichever is smaller.
2861         */
2862        sb->s_maxbytes = MAX_LFS_FILESIZE;
2863        /* Ntfs measures time in 100ns intervals. */
2864        sb->s_time_gran = 100;
2865        /*
2866         * Now load the metadata required for the page cache and our address
2867         * space operations to function. We do this by setting up a specialised
2868         * read_inode method and then just calling the normal iget() to obtain
2869         * the inode for $MFT which is sufficient to allow our normal inode
2870         * operations and associated address space operations to function.
2871         */
2872        sb->s_op = &ntfs_sops;
2873        tmp_ino = new_inode(sb);
2874        if (!tmp_ino) {
2875                if (!silent)
2876                        ntfs_error(sb, "Failed to load essential metadata.");
2877                goto err_out_now;
2878        }
2879        tmp_ino->i_ino = FILE_MFT;
2880        insert_inode_hash(tmp_ino);
2881        if (ntfs_read_inode_mount(tmp_ino) < 0) {
2882                if (!silent)
2883                        ntfs_error(sb, "Failed to load essential metadata.");
2884                goto iput_tmp_ino_err_out_now;
2885        }
2886        mutex_lock(&ntfs_lock);
2887        /*
2888         * The current mount is a compression user if the cluster size is
2889         * less than or equal 4kiB.
2890         */
2891        if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2892                result = allocate_compression_buffers();
2893                if (result) {
2894                        ntfs_error(NULL, "Failed to allocate buffers "
2895                                        "for compression engine.");
2896                        ntfs_nr_compression_users--;
2897                        mutex_unlock(&ntfs_lock);
2898                        goto iput_tmp_ino_err_out_now;
2899                }
2900        }
2901        /*
2902         * Generate the global default upcase table if necessary.  Also
2903         * temporarily increment the number of upcase users to avoid race
2904         * conditions with concurrent (u)mounts.
2905         */
2906        if (!default_upcase)
2907                default_upcase = generate_default_upcase();
2908        ntfs_nr_upcase_users++;
2909        mutex_unlock(&ntfs_lock);
2910        /*
2911         * From now on, ignore @silent parameter. If we fail below this line,
2912         * it will be due to a corrupt fs or a system error, so we report it.
2913         */
2914        /*
2915         * Open the system files with normal access functions and complete
2916         * setting up the ntfs super block.
2917         */
2918        if (!load_system_files(vol)) {
2919                ntfs_error(sb, "Failed to load system files.");
2920                goto unl_upcase_iput_tmp_ino_err_out_now;
2921        }
2922        if ((sb->s_root = d_alloc_root(vol->root_ino))) {
2923                /* We increment i_count simulating an ntfs_iget(). */
2924                atomic_inc(&vol->root_ino->i_count);
2925                ntfs_debug("Exiting, status successful.");
2926                /* Release the default upcase if it has no users. */
2927                mutex_lock(&ntfs_lock);
2928                if (!--ntfs_nr_upcase_users && default_upcase) {
2929                        ntfs_free(default_upcase);
2930                        default_upcase = NULL;
2931                }
2932                mutex_unlock(&ntfs_lock);
2933                sb->s_export_op = &ntfs_export_ops;
2934                lock_kernel();
2935                lockdep_on();
2936                return 0;
2937        }
2938        ntfs_error(sb, "Failed to allocate root directory.");
2939        /* Clean up after the successful load_system_files() call from above. */
2940        // TODO: Use ntfs_put_super() instead of repeating all this code...
2941        // FIXME: Should mark the volume clean as the error is most likely
2942        //        -ENOMEM.
2943        iput(vol->vol_ino);
2944        vol->vol_ino = NULL;
2945        /* NTFS 3.0+ specific clean up. */
2946        if (vol->major_ver >= 3) {
2947#ifdef NTFS_RW
2948                if (vol->usnjrnl_j_ino) {
2949                        iput(vol->usnjrnl_j_ino);
2950                        vol->usnjrnl_j_ino = NULL;
2951                }
2952                if (vol->usnjrnl_max_ino) {
2953                        iput(vol->usnjrnl_max_ino);
2954                        vol->usnjrnl_max_ino = NULL;
2955                }
2956                if (vol->usnjrnl_ino) {
2957                        iput(vol->usnjrnl_ino);
2958                        vol->usnjrnl_ino = NULL;
2959                }
2960                if (vol->quota_q_ino) {
2961                        iput(vol->quota_q_ino);
2962                        vol->quota_q_ino = NULL;
2963                }
2964                if (vol->quota_ino) {
2965                        iput(vol->quota_ino);
2966                        vol->quota_ino = NULL;
2967                }
2968#endif /* NTFS_RW */
2969                if (vol->extend_ino) {
2970                        iput(vol->extend_ino);
2971                        vol->extend_ino = NULL;
2972                }
2973                if (vol->secure_ino) {
2974                        iput(vol->secure_ino);
2975                        vol->secure_ino = NULL;
2976                }
2977        }
2978        iput(vol->root_ino);
2979        vol->root_ino = NULL;
2980        iput(vol->lcnbmp_ino);
2981        vol->lcnbmp_ino = NULL;
2982        iput(vol->mftbmp_ino);
2983        vol->mftbmp_ino = NULL;
2984#ifdef NTFS_RW
2985        if (vol->logfile_ino) {
2986                iput(vol->logfile_ino);
2987                vol->logfile_ino = NULL;
2988        }
2989        if (vol->mftmirr_ino) {
2990                iput(vol->mftmirr_ino);
2991                vol->mftmirr_ino = NULL;
2992        }
2993#endif /* NTFS_RW */
2994        /* Throw away the table of attribute definitions. */
2995        vol->attrdef_size = 0;
2996        if (vol->attrdef) {
2997                ntfs_free(vol->attrdef);
2998                vol->attrdef = NULL;
2999        }
3000        vol->upcase_len = 0;
3001        mutex_lock(&ntfs_lock);
3002        if (vol->upcase == default_upcase) {
3003                ntfs_nr_upcase_users--;
3004                vol->upcase = NULL;
3005        }
3006        mutex_unlock(&ntfs_lock);
3007        if (vol->upcase) {
3008                ntfs_free(vol->upcase);
3009                vol->upcase = NULL;
3010        }
3011        if (vol->nls_map) {
3012                unload_nls(vol->nls_map);
3013                vol->nls_map = NULL;
3014        }
3015        /* Error exit code path. */
3016unl_upcase_iput_tmp_ino_err_out_now:
3017        /*
3018         * Decrease the number of upcase users and destroy the global default
3019         * upcase table if necessary.
3020         */
3021        mutex_lock(&ntfs_lock);
3022        if (!--ntfs_nr_upcase_users && default_upcase) {
3023                ntfs_free(default_upcase);
3024                default_upcase = NULL;
3025        }
3026        if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
3027                free_compression_buffers();
3028        mutex_unlock(&ntfs_lock);
3029iput_tmp_ino_err_out_now:
3030        iput(tmp_ino);
3031        if (vol->mft_ino && vol->mft_ino != tmp_ino)
3032                iput(vol->mft_ino);
3033        vol->mft_ino = NULL;
3034        /*
3035         * This is needed to get ntfs_clear_extent_inode() called for each
3036         * inode we have ever called ntfs_iget()/iput() on, otherwise we A)
3037         * leak resources and B) a subsequent mount fails automatically due to
3038         * ntfs_iget() never calling down into our ntfs_read_locked_inode()
3039         * method again... FIXME: Do we need to do this twice now because of
3040         * attribute inodes? I think not, so leave as is for now... (AIA)
3041         */
3042        if (invalidate_inodes(sb)) {
3043                ntfs_error(sb, "Busy inodes left. This is most likely a NTFS "
3044                                "driver bug.");
3045                /* Copied from fs/super.c. I just love this message. (-; */
3046                printk("NTFS: Busy inodes after umount. Self-destruct in 5 "
3047                                "seconds.  Have a nice day...\n");
3048        }
3049        /* Errors at this stage are irrelevant. */
3050err_out_now:
3051        lock_kernel();
3052        sb->s_fs_info = NULL;
3053        kfree(vol);
3054        ntfs_debug("Failed, returning -EINVAL.");
3055        lockdep_on();
3056        return -EINVAL;
3057}
3058
3059/*
3060 * This is a slab cache to optimize allocations and deallocations of Unicode
3061 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3062 * (255) Unicode characters + a terminating NULL Unicode character.
3063 */
3064struct kmem_cache *ntfs_name_cache;
3065
3066/* Slab caches for efficient allocation/deallocation of inodes. */
3067struct kmem_cache *ntfs_inode_cache;
3068struct kmem_cache *ntfs_big_inode_cache;
3069
3070/* Init once constructor for the inode slab cache. */
3071static void ntfs_big_inode_init_once(void *foo)
3072{
3073        ntfs_inode *ni = (ntfs_inode *)foo;
3074
3075        inode_init_once(VFS_I(ni));
3076}
3077
3078/*
3079 * Slab caches to optimize allocations and deallocations of attribute search
3080 * contexts and index contexts, respectively.
3081 */
3082struct kmem_cache *ntfs_attr_ctx_cache;
3083struct kmem_cache *ntfs_index_ctx_cache;
3084
3085/* Driver wide mutex. */
3086DEFINE_MUTEX(ntfs_lock);
3087
3088static int ntfs_get_sb(struct file_system_type *fs_type,
3089        int flags, const char *dev_name, void *data, struct vfsmount *mnt)
3090{
3091        return get_sb_bdev(fs_type, flags, dev_name, data, ntfs_fill_super,
3092                           mnt);
3093}
3094
3095static struct file_system_type ntfs_fs_type = {
3096        .owner          = THIS_MODULE,
3097        .name           = "ntfs",
3098        .get_sb         = ntfs_get_sb,
3099        .kill_sb        = kill_block_super,
3100        .fs_flags       = FS_REQUIRES_DEV,
3101};
3102
3103/* Stable names for the slab caches. */
3104static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3105static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3106static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3107static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3108static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3109
3110static int __init init_ntfs_fs(void)
3111{
3112        int err = 0;
3113
3114        /* This may be ugly but it results in pretty output so who cares. (-8 */
3115        printk(KERN_INFO "NTFS driver " NTFS_VERSION " [Flags: R/"
3116#ifdef NTFS_RW
3117                        "W"
3118#else
3119                        "O"
3120#endif
3121#ifdef DEBUG
3122                        " DEBUG"
3123#endif
3124#ifdef MODULE
3125                        " MODULE"
3126#endif
3127                        "].\n");
3128
3129        ntfs_debug("Debug messages are enabled.");
3130
3131        ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3132                        sizeof(ntfs_index_context), 0 /* offset */,
3133                        SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3134        if (!ntfs_index_ctx_cache) {
3135                printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3136                                ntfs_index_ctx_cache_name);
3137                goto ictx_err_out;
3138        }
3139        ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3140                        sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3141                        SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3142        if (!ntfs_attr_ctx_cache) {
3143                printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3144                                ntfs_attr_ctx_cache_name);
3145                goto actx_err_out;
3146        }
3147
3148        ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3149                        (NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3150                        SLAB_HWCACHE_ALIGN, NULL);
3151        if (!ntfs_name_cache) {
3152                printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3153                                ntfs_name_cache_name);
3154                goto name_err_out;
3155        }
3156
3157        ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3158                        sizeof(ntfs_inode), 0,
3159                        SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
3160        if (!ntfs_inode_cache) {
3161                printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3162                                ntfs_inode_cache_name);
3163                goto inode_err_out;
3164        }
3165
3166        ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3167                        sizeof(big_ntfs_inode), 0,
3168                        SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
3169                        ntfs_big_inode_init_once);
3170        if (!ntfs_big_inode_cache) {
3171                printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3172                                ntfs_big_inode_cache_name);
3173                goto big_inode_err_out;
3174        }
3175
3176        /* Register the ntfs sysctls. */
3177        err = ntfs_sysctl(1);
3178        if (err) {
3179                printk(KERN_CRIT "NTFS: Failed to register NTFS sysctls!\n");
3180                goto sysctl_err_out;
3181        }
3182
3183        err = register_filesystem(&ntfs_fs_type);
3184        if (!err) {
3185                ntfs_debug("NTFS driver registered successfully.");
3186                return 0; /* Success! */
3187        }
3188        printk(KERN_CRIT "NTFS: Failed to register NTFS filesystem driver!\n");
3189
3190sysctl_err_out:
3191        kmem_cache_destroy(ntfs_big_inode_cache);
3192big_inode_err_out:
3193        kmem_cache_destroy(ntfs_inode_cache);
3194inode_err_out:
3195        kmem_cache_destroy(ntfs_name_cache);
3196name_err_out:
3197        kmem_cache_destroy(ntfs_attr_ctx_cache);
3198actx_err_out:
3199        kmem_cache_destroy(ntfs_index_ctx_cache);
3200ictx_err_out:
3201        if (!err) {
3202                printk(KERN_CRIT "NTFS: Aborting NTFS filesystem driver "
3203                                "registration...\n");
3204                err = -ENOMEM;
3205        }
3206        return err;
3207}
3208
3209static void __exit exit_ntfs_fs(void)
3210{
3211        ntfs_debug("Unregistering NTFS driver.");
3212
3213        unregister_filesystem(&ntfs_fs_type);
3214        kmem_cache_destroy(ntfs_big_inode_cache);
3215        kmem_cache_destroy(ntfs_inode_cache);
3216        kmem_cache_destroy(ntfs_name_cache);
3217        kmem_cache_destroy(ntfs_attr_ctx_cache);
3218        kmem_cache_destroy(ntfs_index_ctx_cache);
3219        /* Unregister the ntfs sysctls. */
3220        ntfs_sysctl(0);
3221}
3222
3223MODULE_AUTHOR("Anton Altaparmakov <aia21@cantab.net>");
3224MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2007 Anton Altaparmakov");
3225MODULE_VERSION(NTFS_VERSION);
3226MODULE_LICENSE("GPL");
3227#ifdef DEBUG
3228module_param(debug_msgs, bool, 0);
3229MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3230#endif
3231
3232module_init(init_ntfs_fs)
3233module_exit(exit_ntfs_fs)
3234