cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

super.c (101010B)


      1// SPDX-License-Identifier: GPL-2.0-or-later
      2/*
      3 * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
      4 *
      5 * Copyright (c) 2001-2012 Anton Altaparmakov and Tuxera Inc.
      6 * Copyright (c) 2001,2002 Richard Russon
      7 */
      8#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
      9
     10#include <linux/stddef.h>
     11#include <linux/init.h>
     12#include <linux/slab.h>
     13#include <linux/string.h>
     14#include <linux/spinlock.h>
     15#include <linux/blkdev.h>	/* For bdev_logical_block_size(). */
     16#include <linux/backing-dev.h>
     17#include <linux/buffer_head.h>
     18#include <linux/vfs.h>
     19#include <linux/moduleparam.h>
     20#include <linux/bitmap.h>
     21
     22#include "sysctl.h"
     23#include "logfile.h"
     24#include "quota.h"
     25#include "usnjrnl.h"
     26#include "dir.h"
     27#include "debug.h"
     28#include "index.h"
     29#include "inode.h"
     30#include "aops.h"
     31#include "layout.h"
     32#include "malloc.h"
     33#include "ntfs.h"
     34
     35/* Number of mounted filesystems which have compression enabled. */
     36static unsigned long ntfs_nr_compression_users;
     37
     38/* A global default upcase table and a corresponding reference count. */
     39static ntfschar *default_upcase;
     40static unsigned long ntfs_nr_upcase_users;
     41
     42/* Error constants/strings used in inode.c::ntfs_show_options(). */
     43typedef enum {
     44	/* One of these must be present, default is ON_ERRORS_CONTINUE. */
     45	ON_ERRORS_PANIC			= 0x01,
     46	ON_ERRORS_REMOUNT_RO		= 0x02,
     47	ON_ERRORS_CONTINUE		= 0x04,
     48	/* Optional, can be combined with any of the above. */
     49	ON_ERRORS_RECOVER		= 0x10,
     50} ON_ERRORS_ACTIONS;
     51
     52const option_t on_errors_arr[] = {
     53	{ ON_ERRORS_PANIC,	"panic" },
     54	{ ON_ERRORS_REMOUNT_RO,	"remount-ro", },
     55	{ ON_ERRORS_CONTINUE,	"continue", },
     56	{ ON_ERRORS_RECOVER,	"recover" },
     57	{ 0,			NULL }
     58};
     59
     60/**
     61 * simple_getbool -
     62 *
     63 * Copied from old ntfs driver (which copied from vfat driver).
     64 */
     65static int simple_getbool(char *s, bool *setval)
     66{
     67	if (s) {
     68		if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
     69			*setval = true;
     70		else if (!strcmp(s, "0") || !strcmp(s, "no") ||
     71							!strcmp(s, "false"))
     72			*setval = false;
     73		else
     74			return 0;
     75	} else
     76		*setval = true;
     77	return 1;
     78}
     79
     80/**
     81 * parse_options - parse the (re)mount options
     82 * @vol:	ntfs volume
     83 * @opt:	string containing the (re)mount options
     84 *
     85 * Parse the recognized options in @opt for the ntfs volume described by @vol.
     86 */
     87static bool parse_options(ntfs_volume *vol, char *opt)
     88{
     89	char *p, *v, *ov;
     90	static char *utf8 = "utf8";
     91	int errors = 0, sloppy = 0;
     92	kuid_t uid = INVALID_UID;
     93	kgid_t gid = INVALID_GID;
     94	umode_t fmask = (umode_t)-1, dmask = (umode_t)-1;
     95	int mft_zone_multiplier = -1, on_errors = -1;
     96	int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
     97	struct nls_table *nls_map = NULL, *old_nls;
     98
     99	/* I am lazy... (-8 */
    100#define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value)	\
    101	if (!strcmp(p, option)) {					\
    102		if (!v || !*v)						\
    103			variable = default_value;			\
    104		else {							\
    105			variable = simple_strtoul(ov = v, &v, 0);	\
    106			if (*v)						\
    107				goto needs_val;				\
    108		}							\
    109	}
    110#define NTFS_GETOPT(option, variable)					\
    111	if (!strcmp(p, option)) {					\
    112		if (!v || !*v)						\
    113			goto needs_arg;					\
    114		variable = simple_strtoul(ov = v, &v, 0);		\
    115		if (*v)							\
    116			goto needs_val;					\
    117	}
    118#define NTFS_GETOPT_UID(option, variable)				\
    119	if (!strcmp(p, option)) {					\
    120		uid_t uid_value;					\
    121		if (!v || !*v)						\
    122			goto needs_arg;					\
    123		uid_value = simple_strtoul(ov = v, &v, 0);		\
    124		if (*v)							\
    125			goto needs_val;					\
    126		variable = make_kuid(current_user_ns(), uid_value);	\
    127		if (!uid_valid(variable))				\
    128			goto needs_val;					\
    129	}
    130#define NTFS_GETOPT_GID(option, variable)				\
    131	if (!strcmp(p, option)) {					\
    132		gid_t gid_value;					\
    133		if (!v || !*v)						\
    134			goto needs_arg;					\
    135		gid_value = simple_strtoul(ov = v, &v, 0);		\
    136		if (*v)							\
    137			goto needs_val;					\
    138		variable = make_kgid(current_user_ns(), gid_value);	\
    139		if (!gid_valid(variable))				\
    140			goto needs_val;					\
    141	}
    142#define NTFS_GETOPT_OCTAL(option, variable)				\
    143	if (!strcmp(p, option)) {					\
    144		if (!v || !*v)						\
    145			goto needs_arg;					\
    146		variable = simple_strtoul(ov = v, &v, 8);		\
    147		if (*v)							\
    148			goto needs_val;					\
    149	}
    150#define NTFS_GETOPT_BOOL(option, variable)				\
    151	if (!strcmp(p, option)) {					\
    152		bool val;						\
    153		if (!simple_getbool(v, &val))				\
    154			goto needs_bool;				\
    155		variable = val;						\
    156	}
    157#define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array)		\
    158	if (!strcmp(p, option)) {					\
    159		int _i;							\
    160		if (!v || !*v)						\
    161			goto needs_arg;					\
    162		ov = v;							\
    163		if (variable == -1)					\
    164			variable = 0;					\
    165		for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
    166			if (!strcmp(opt_array[_i].str, v)) {		\
    167				variable |= opt_array[_i].val;		\
    168				break;					\
    169			}						\
    170		if (!opt_array[_i].str || !*opt_array[_i].str)		\
    171			goto needs_val;					\
    172	}
    173	if (!opt || !*opt)
    174		goto no_mount_options;
    175	ntfs_debug("Entering with mount options string: %s", opt);
    176	while ((p = strsep(&opt, ","))) {
    177		if ((v = strchr(p, '=')))
    178			*v++ = 0;
    179		NTFS_GETOPT_UID("uid", uid)
    180		else NTFS_GETOPT_GID("gid", gid)
    181		else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
    182		else NTFS_GETOPT_OCTAL("fmask", fmask)
    183		else NTFS_GETOPT_OCTAL("dmask", dmask)
    184		else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
    185		else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
    186		else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
    187		else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
    188		else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
    189		else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
    190				on_errors_arr)
    191		else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
    192			ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
    193					p);
    194		else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
    195			if (!strcmp(p, "iocharset"))
    196				ntfs_warning(vol->sb, "Option iocharset is "
    197						"deprecated. Please use "
    198						"option nls=<charsetname> in "
    199						"the future.");
    200			if (!v || !*v)
    201				goto needs_arg;
    202use_utf8:
    203			old_nls = nls_map;
    204			nls_map = load_nls(v);
    205			if (!nls_map) {
    206				if (!old_nls) {
    207					ntfs_error(vol->sb, "NLS character set "
    208							"%s not found.", v);
    209					return false;
    210				}
    211				ntfs_error(vol->sb, "NLS character set %s not "
    212						"found. Using previous one %s.",
    213						v, old_nls->charset);
    214				nls_map = old_nls;
    215			} else /* nls_map */ {
    216				unload_nls(old_nls);
    217			}
    218		} else if (!strcmp(p, "utf8")) {
    219			bool val = false;
    220			ntfs_warning(vol->sb, "Option utf8 is no longer "
    221				   "supported, using option nls=utf8. Please "
    222				   "use option nls=utf8 in the future and "
    223				   "make sure utf8 is compiled either as a "
    224				   "module or into the kernel.");
    225			if (!v || !*v)
    226				val = true;
    227			else if (!simple_getbool(v, &val))
    228				goto needs_bool;
    229			if (val) {
    230				v = utf8;
    231				goto use_utf8;
    232			}
    233		} else {
    234			ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
    235			if (errors < INT_MAX)
    236				errors++;
    237		}
    238#undef NTFS_GETOPT_OPTIONS_ARRAY
    239#undef NTFS_GETOPT_BOOL
    240#undef NTFS_GETOPT
    241#undef NTFS_GETOPT_WITH_DEFAULT
    242	}
    243no_mount_options:
    244	if (errors && !sloppy)
    245		return false;
    246	if (sloppy)
    247		ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
    248				"unrecognized mount option(s) and continuing.");
    249	/* Keep this first! */
    250	if (on_errors != -1) {
    251		if (!on_errors) {
    252			ntfs_error(vol->sb, "Invalid errors option argument "
    253					"or bug in options parser.");
    254			return false;
    255		}
    256	}
    257	if (nls_map) {
    258		if (vol->nls_map && vol->nls_map != nls_map) {
    259			ntfs_error(vol->sb, "Cannot change NLS character set "
    260					"on remount.");
    261			return false;
    262		} /* else (!vol->nls_map) */
    263		ntfs_debug("Using NLS character set %s.", nls_map->charset);
    264		vol->nls_map = nls_map;
    265	} else /* (!nls_map) */ {
    266		if (!vol->nls_map) {
    267			vol->nls_map = load_nls_default();
    268			if (!vol->nls_map) {
    269				ntfs_error(vol->sb, "Failed to load default "
    270						"NLS character set.");
    271				return false;
    272			}
    273			ntfs_debug("Using default NLS character set (%s).",
    274					vol->nls_map->charset);
    275		}
    276	}
    277	if (mft_zone_multiplier != -1) {
    278		if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
    279				mft_zone_multiplier) {
    280			ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
    281					"on remount.");
    282			return false;
    283		}
    284		if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
    285			ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
    286					"Using default value, i.e. 1.");
    287			mft_zone_multiplier = 1;
    288		}
    289		vol->mft_zone_multiplier = mft_zone_multiplier;
    290	}
    291	if (!vol->mft_zone_multiplier)
    292		vol->mft_zone_multiplier = 1;
    293	if (on_errors != -1)
    294		vol->on_errors = on_errors;
    295	if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
    296		vol->on_errors |= ON_ERRORS_CONTINUE;
    297	if (uid_valid(uid))
    298		vol->uid = uid;
    299	if (gid_valid(gid))
    300		vol->gid = gid;
    301	if (fmask != (umode_t)-1)
    302		vol->fmask = fmask;
    303	if (dmask != (umode_t)-1)
    304		vol->dmask = dmask;
    305	if (show_sys_files != -1) {
    306		if (show_sys_files)
    307			NVolSetShowSystemFiles(vol);
    308		else
    309			NVolClearShowSystemFiles(vol);
    310	}
    311	if (case_sensitive != -1) {
    312		if (case_sensitive)
    313			NVolSetCaseSensitive(vol);
    314		else
    315			NVolClearCaseSensitive(vol);
    316	}
    317	if (disable_sparse != -1) {
    318		if (disable_sparse)
    319			NVolClearSparseEnabled(vol);
    320		else {
    321			if (!NVolSparseEnabled(vol) &&
    322					vol->major_ver && vol->major_ver < 3)
    323				ntfs_warning(vol->sb, "Not enabling sparse "
    324						"support due to NTFS volume "
    325						"version %i.%i (need at least "
    326						"version 3.0).", vol->major_ver,
    327						vol->minor_ver);
    328			else
    329				NVolSetSparseEnabled(vol);
    330		}
    331	}
    332	return true;
    333needs_arg:
    334	ntfs_error(vol->sb, "The %s option requires an argument.", p);
    335	return false;
    336needs_bool:
    337	ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
    338	return false;
    339needs_val:
    340	ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
    341	return false;
    342}
    343
    344#ifdef NTFS_RW
    345
    346/**
    347 * ntfs_write_volume_flags - write new flags to the volume information flags
    348 * @vol:	ntfs volume on which to modify the flags
    349 * @flags:	new flags value for the volume information flags
    350 *
    351 * Internal function.  You probably want to use ntfs_{set,clear}_volume_flags()
    352 * instead (see below).
    353 *
    354 * Replace the volume information flags on the volume @vol with the value
    355 * supplied in @flags.  Note, this overwrites the volume information flags, so
    356 * make sure to combine the flags you want to modify with the old flags and use
    357 * the result when calling ntfs_write_volume_flags().
    358 *
    359 * Return 0 on success and -errno on error.
    360 */
    361static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
    362{
    363	ntfs_inode *ni = NTFS_I(vol->vol_ino);
    364	MFT_RECORD *m;
    365	VOLUME_INFORMATION *vi;
    366	ntfs_attr_search_ctx *ctx;
    367	int err;
    368
    369	ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
    370			le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
    371	if (vol->vol_flags == flags)
    372		goto done;
    373	BUG_ON(!ni);
    374	m = map_mft_record(ni);
    375	if (IS_ERR(m)) {
    376		err = PTR_ERR(m);
    377		goto err_out;
    378	}
    379	ctx = ntfs_attr_get_search_ctx(ni, m);
    380	if (!ctx) {
    381		err = -ENOMEM;
    382		goto put_unm_err_out;
    383	}
    384	err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
    385			ctx);
    386	if (err)
    387		goto put_unm_err_out;
    388	vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
    389			le16_to_cpu(ctx->attr->data.resident.value_offset));
    390	vol->vol_flags = vi->flags = flags;
    391	flush_dcache_mft_record_page(ctx->ntfs_ino);
    392	mark_mft_record_dirty(ctx->ntfs_ino);
    393	ntfs_attr_put_search_ctx(ctx);
    394	unmap_mft_record(ni);
    395done:
    396	ntfs_debug("Done.");
    397	return 0;
    398put_unm_err_out:
    399	if (ctx)
    400		ntfs_attr_put_search_ctx(ctx);
    401	unmap_mft_record(ni);
    402err_out:
    403	ntfs_error(vol->sb, "Failed with error code %i.", -err);
    404	return err;
    405}
    406
    407/**
    408 * ntfs_set_volume_flags - set bits in the volume information flags
    409 * @vol:	ntfs volume on which to modify the flags
    410 * @flags:	flags to set on the volume
    411 *
    412 * Set the bits in @flags in the volume information flags on the volume @vol.
    413 *
    414 * Return 0 on success and -errno on error.
    415 */
    416static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
    417{
    418	flags &= VOLUME_FLAGS_MASK;
    419	return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
    420}
    421
    422/**
    423 * ntfs_clear_volume_flags - clear bits in the volume information flags
    424 * @vol:	ntfs volume on which to modify the flags
    425 * @flags:	flags to clear on the volume
    426 *
    427 * Clear the bits in @flags in the volume information flags on the volume @vol.
    428 *
    429 * Return 0 on success and -errno on error.
    430 */
    431static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
    432{
    433	flags &= VOLUME_FLAGS_MASK;
    434	flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
    435	return ntfs_write_volume_flags(vol, flags);
    436}
    437
    438#endif /* NTFS_RW */
    439
    440/**
    441 * ntfs_remount - change the mount options of a mounted ntfs filesystem
    442 * @sb:		superblock of mounted ntfs filesystem
    443 * @flags:	remount flags
    444 * @opt:	remount options string
    445 *
    446 * Change the mount options of an already mounted ntfs filesystem.
    447 *
    448 * NOTE:  The VFS sets the @sb->s_flags remount flags to @flags after
    449 * ntfs_remount() returns successfully (i.e. returns 0).  Otherwise,
    450 * @sb->s_flags are not changed.
    451 */
    452static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
    453{
    454	ntfs_volume *vol = NTFS_SB(sb);
    455
    456	ntfs_debug("Entering with remount options string: %s", opt);
    457
    458	sync_filesystem(sb);
    459
    460#ifndef NTFS_RW
    461	/* For read-only compiled driver, enforce read-only flag. */
    462	*flags |= SB_RDONLY;
    463#else /* NTFS_RW */
    464	/*
    465	 * For the read-write compiled driver, if we are remounting read-write,
    466	 * make sure there are no volume errors and that no unsupported volume
    467	 * flags are set.  Also, empty the logfile journal as it would become
    468	 * stale as soon as something is written to the volume and mark the
    469	 * volume dirty so that chkdsk is run if the volume is not umounted
    470	 * cleanly.  Finally, mark the quotas out of date so Windows rescans
    471	 * the volume on boot and updates them.
    472	 *
    473	 * When remounting read-only, mark the volume clean if no volume errors
    474	 * have occurred.
    475	 */
    476	if (sb_rdonly(sb) && !(*flags & SB_RDONLY)) {
    477		static const char *es = ".  Cannot remount read-write.";
    478
    479		/* Remounting read-write. */
    480		if (NVolErrors(vol)) {
    481			ntfs_error(sb, "Volume has errors and is read-only%s",
    482					es);
    483			return -EROFS;
    484		}
    485		if (vol->vol_flags & VOLUME_IS_DIRTY) {
    486			ntfs_error(sb, "Volume is dirty and read-only%s", es);
    487			return -EROFS;
    488		}
    489		if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
    490			ntfs_error(sb, "Volume has been modified by chkdsk "
    491					"and is read-only%s", es);
    492			return -EROFS;
    493		}
    494		if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
    495			ntfs_error(sb, "Volume has unsupported flags set "
    496					"(0x%x) and is read-only%s",
    497					(unsigned)le16_to_cpu(vol->vol_flags),
    498					es);
    499			return -EROFS;
    500		}
    501		if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
    502			ntfs_error(sb, "Failed to set dirty bit in volume "
    503					"information flags%s", es);
    504			return -EROFS;
    505		}
    506#if 0
    507		// TODO: Enable this code once we start modifying anything that
    508		//	 is different between NTFS 1.2 and 3.x...
    509		/* Set NT4 compatibility flag on newer NTFS version volumes. */
    510		if ((vol->major_ver > 1)) {
    511			if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
    512				ntfs_error(sb, "Failed to set NT4 "
    513						"compatibility flag%s", es);
    514				NVolSetErrors(vol);
    515				return -EROFS;
    516			}
    517		}
    518#endif
    519		if (!ntfs_empty_logfile(vol->logfile_ino)) {
    520			ntfs_error(sb, "Failed to empty journal $LogFile%s",
    521					es);
    522			NVolSetErrors(vol);
    523			return -EROFS;
    524		}
    525		if (!ntfs_mark_quotas_out_of_date(vol)) {
    526			ntfs_error(sb, "Failed to mark quotas out of date%s",
    527					es);
    528			NVolSetErrors(vol);
    529			return -EROFS;
    530		}
    531		if (!ntfs_stamp_usnjrnl(vol)) {
    532			ntfs_error(sb, "Failed to stamp transaction log "
    533					"($UsnJrnl)%s", es);
    534			NVolSetErrors(vol);
    535			return -EROFS;
    536		}
    537	} else if (!sb_rdonly(sb) && (*flags & SB_RDONLY)) {
    538		/* Remounting read-only. */
    539		if (!NVolErrors(vol)) {
    540			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
    541				ntfs_warning(sb, "Failed to clear dirty bit "
    542						"in volume information "
    543						"flags.  Run chkdsk.");
    544		}
    545	}
    546#endif /* NTFS_RW */
    547
    548	// TODO: Deal with *flags.
    549
    550	if (!parse_options(vol, opt))
    551		return -EINVAL;
    552
    553	ntfs_debug("Done.");
    554	return 0;
    555}
    556
    557/**
    558 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
    559 * @sb:		Super block of the device to which @b belongs.
    560 * @b:		Boot sector of device @sb to check.
    561 * @silent:	If 'true', all output will be silenced.
    562 *
    563 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
    564 * sector. Returns 'true' if it is valid and 'false' if not.
    565 *
    566 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
    567 * is 'true'.
    568 */
    569static bool is_boot_sector_ntfs(const struct super_block *sb,
    570		const NTFS_BOOT_SECTOR *b, const bool silent)
    571{
    572	/*
    573	 * Check that checksum == sum of u32 values from b to the checksum
    574	 * field.  If checksum is zero, no checking is done.  We will work when
    575	 * the checksum test fails, since some utilities update the boot sector
    576	 * ignoring the checksum which leaves the checksum out-of-date.  We
    577	 * report a warning if this is the case.
    578	 */
    579	if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
    580		le32 *u;
    581		u32 i;
    582
    583		for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
    584			i += le32_to_cpup(u);
    585		if (le32_to_cpu(b->checksum) != i)
    586			ntfs_warning(sb, "Invalid boot sector checksum.");
    587	}
    588	/* Check OEMidentifier is "NTFS    " */
    589	if (b->oem_id != magicNTFS)
    590		goto not_ntfs;
    591	/* Check bytes per sector value is between 256 and 4096. */
    592	if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
    593			le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
    594		goto not_ntfs;
    595	/* Check sectors per cluster value is valid. */
    596	switch (b->bpb.sectors_per_cluster) {
    597	case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
    598		break;
    599	default:
    600		goto not_ntfs;
    601	}
    602	/* Check the cluster size is not above the maximum (64kiB). */
    603	if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
    604			b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
    605		goto not_ntfs;
    606	/* Check reserved/unused fields are really zero. */
    607	if (le16_to_cpu(b->bpb.reserved_sectors) ||
    608			le16_to_cpu(b->bpb.root_entries) ||
    609			le16_to_cpu(b->bpb.sectors) ||
    610			le16_to_cpu(b->bpb.sectors_per_fat) ||
    611			le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
    612		goto not_ntfs;
    613	/* Check clusters per file mft record value is valid. */
    614	if ((u8)b->clusters_per_mft_record < 0xe1 ||
    615			(u8)b->clusters_per_mft_record > 0xf7)
    616		switch (b->clusters_per_mft_record) {
    617		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
    618			break;
    619		default:
    620			goto not_ntfs;
    621		}
    622	/* Check clusters per index block value is valid. */
    623	if ((u8)b->clusters_per_index_record < 0xe1 ||
    624			(u8)b->clusters_per_index_record > 0xf7)
    625		switch (b->clusters_per_index_record) {
    626		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
    627			break;
    628		default:
    629			goto not_ntfs;
    630		}
    631	/*
    632	 * Check for valid end of sector marker. We will work without it, but
    633	 * many BIOSes will refuse to boot from a bootsector if the magic is
    634	 * incorrect, so we emit a warning.
    635	 */
    636	if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
    637		ntfs_warning(sb, "Invalid end of sector marker.");
    638	return true;
    639not_ntfs:
    640	return false;
    641}
    642
    643/**
    644 * read_ntfs_boot_sector - read the NTFS boot sector of a device
    645 * @sb:		super block of device to read the boot sector from
    646 * @silent:	if true, suppress all output
    647 *
    648 * Reads the boot sector from the device and validates it. If that fails, tries
    649 * to read the backup boot sector, first from the end of the device a-la NT4 and
    650 * later and then from the middle of the device a-la NT3.51 and before.
    651 *
    652 * If a valid boot sector is found but it is not the primary boot sector, we
    653 * repair the primary boot sector silently (unless the device is read-only or
    654 * the primary boot sector is not accessible).
    655 *
    656 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
    657 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
    658 * to their respective values.
    659 *
    660 * Return the unlocked buffer head containing the boot sector or NULL on error.
    661 */
    662static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
    663		const int silent)
    664{
    665	const char *read_err_str = "Unable to read %s boot sector.";
    666	struct buffer_head *bh_primary, *bh_backup;
    667	sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
    668
    669	/* Try to read primary boot sector. */
    670	if ((bh_primary = sb_bread(sb, 0))) {
    671		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
    672				bh_primary->b_data, silent))
    673			return bh_primary;
    674		if (!silent)
    675			ntfs_error(sb, "Primary boot sector is invalid.");
    676	} else if (!silent)
    677		ntfs_error(sb, read_err_str, "primary");
    678	if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
    679		if (bh_primary)
    680			brelse(bh_primary);
    681		if (!silent)
    682			ntfs_error(sb, "Mount option errors=recover not used. "
    683					"Aborting without trying to recover.");
    684		return NULL;
    685	}
    686	/* Try to read NT4+ backup boot sector. */
    687	if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
    688		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
    689				bh_backup->b_data, silent))
    690			goto hotfix_primary_boot_sector;
    691		brelse(bh_backup);
    692	} else if (!silent)
    693		ntfs_error(sb, read_err_str, "backup");
    694	/* Try to read NT3.51- backup boot sector. */
    695	if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
    696		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
    697				bh_backup->b_data, silent))
    698			goto hotfix_primary_boot_sector;
    699		if (!silent)
    700			ntfs_error(sb, "Could not find a valid backup boot "
    701					"sector.");
    702		brelse(bh_backup);
    703	} else if (!silent)
    704		ntfs_error(sb, read_err_str, "backup");
    705	/* We failed. Cleanup and return. */
    706	if (bh_primary)
    707		brelse(bh_primary);
    708	return NULL;
    709hotfix_primary_boot_sector:
    710	if (bh_primary) {
    711		/*
    712		 * If we managed to read sector zero and the volume is not
    713		 * read-only, copy the found, valid backup boot sector to the
    714		 * primary boot sector.  Note we only copy the actual boot
    715		 * sector structure, not the actual whole device sector as that
    716		 * may be bigger and would potentially damage the $Boot system
    717		 * file (FIXME: Would be nice to know if the backup boot sector
    718		 * on a large sector device contains the whole boot loader or
    719		 * just the first 512 bytes).
    720		 */
    721		if (!sb_rdonly(sb)) {
    722			ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
    723					"boot sector from backup copy.");
    724			memcpy(bh_primary->b_data, bh_backup->b_data,
    725					NTFS_BLOCK_SIZE);
    726			mark_buffer_dirty(bh_primary);
    727			sync_dirty_buffer(bh_primary);
    728			if (buffer_uptodate(bh_primary)) {
    729				brelse(bh_backup);
    730				return bh_primary;
    731			}
    732			ntfs_error(sb, "Hot-fix: Device write error while "
    733					"recovering primary boot sector.");
    734		} else {
    735			ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
    736					"sector failed: Read-only mount.");
    737		}
    738		brelse(bh_primary);
    739	}
    740	ntfs_warning(sb, "Using backup boot sector.");
    741	return bh_backup;
    742}
    743
    744/**
    745 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
    746 * @vol:	volume structure to initialise with data from boot sector
    747 * @b:		boot sector to parse
    748 *
    749 * Parse the ntfs boot sector @b and store all imporant information therein in
    750 * the ntfs super block @vol.  Return 'true' on success and 'false' on error.
    751 */
    752static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
    753{
    754	unsigned int sectors_per_cluster_bits, nr_hidden_sects;
    755	int clusters_per_mft_record, clusters_per_index_record;
    756	s64 ll;
    757
    758	vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
    759	vol->sector_size_bits = ffs(vol->sector_size) - 1;
    760	ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
    761			vol->sector_size);
    762	ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
    763			vol->sector_size_bits);
    764	if (vol->sector_size < vol->sb->s_blocksize) {
    765		ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
    766				"device block size (%lu).  This is not "
    767				"supported.  Sorry.", vol->sector_size,
    768				vol->sb->s_blocksize);
    769		return false;
    770	}
    771	ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
    772	sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
    773	ntfs_debug("sectors_per_cluster_bits = 0x%x",
    774			sectors_per_cluster_bits);
    775	nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
    776	ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
    777	vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
    778	vol->cluster_size_mask = vol->cluster_size - 1;
    779	vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
    780	ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
    781			vol->cluster_size);
    782	ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
    783	ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
    784	if (vol->cluster_size < vol->sector_size) {
    785		ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
    786				"sector size (%i).  This is not supported.  "
    787				"Sorry.", vol->cluster_size, vol->sector_size);
    788		return false;
    789	}
    790	clusters_per_mft_record = b->clusters_per_mft_record;
    791	ntfs_debug("clusters_per_mft_record = %i (0x%x)",
    792			clusters_per_mft_record, clusters_per_mft_record);
    793	if (clusters_per_mft_record > 0)
    794		vol->mft_record_size = vol->cluster_size <<
    795				(ffs(clusters_per_mft_record) - 1);
    796	else
    797		/*
    798		 * When mft_record_size < cluster_size, clusters_per_mft_record
    799		 * = -log2(mft_record_size) bytes. mft_record_size normaly is
    800		 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
    801		 */
    802		vol->mft_record_size = 1 << -clusters_per_mft_record;
    803	vol->mft_record_size_mask = vol->mft_record_size - 1;
    804	vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
    805	ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
    806			vol->mft_record_size);
    807	ntfs_debug("vol->mft_record_size_mask = 0x%x",
    808			vol->mft_record_size_mask);
    809	ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
    810			vol->mft_record_size_bits, vol->mft_record_size_bits);
    811	/*
    812	 * We cannot support mft record sizes above the PAGE_SIZE since
    813	 * we store $MFT/$DATA, the table of mft records in the page cache.
    814	 */
    815	if (vol->mft_record_size > PAGE_SIZE) {
    816		ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
    817				"PAGE_SIZE on your system (%lu).  "
    818				"This is not supported.  Sorry.",
    819				vol->mft_record_size, PAGE_SIZE);
    820		return false;
    821	}
    822	/* We cannot support mft record sizes below the sector size. */
    823	if (vol->mft_record_size < vol->sector_size) {
    824		ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
    825				"sector size (%i).  This is not supported.  "
    826				"Sorry.", vol->mft_record_size,
    827				vol->sector_size);
    828		return false;
    829	}
    830	clusters_per_index_record = b->clusters_per_index_record;
    831	ntfs_debug("clusters_per_index_record = %i (0x%x)",
    832			clusters_per_index_record, clusters_per_index_record);
    833	if (clusters_per_index_record > 0)
    834		vol->index_record_size = vol->cluster_size <<
    835				(ffs(clusters_per_index_record) - 1);
    836	else
    837		/*
    838		 * When index_record_size < cluster_size,
    839		 * clusters_per_index_record = -log2(index_record_size) bytes.
    840		 * index_record_size normaly equals 4096 bytes, which is
    841		 * encoded as 0xF4 (-12 in decimal).
    842		 */
    843		vol->index_record_size = 1 << -clusters_per_index_record;
    844	vol->index_record_size_mask = vol->index_record_size - 1;
    845	vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
    846	ntfs_debug("vol->index_record_size = %i (0x%x)",
    847			vol->index_record_size, vol->index_record_size);
    848	ntfs_debug("vol->index_record_size_mask = 0x%x",
    849			vol->index_record_size_mask);
    850	ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
    851			vol->index_record_size_bits,
    852			vol->index_record_size_bits);
    853	/* We cannot support index record sizes below the sector size. */
    854	if (vol->index_record_size < vol->sector_size) {
    855		ntfs_error(vol->sb, "Index record size (%i) is smaller than "
    856				"the sector size (%i).  This is not "
    857				"supported.  Sorry.", vol->index_record_size,
    858				vol->sector_size);
    859		return false;
    860	}
    861	/*
    862	 * Get the size of the volume in clusters and check for 64-bit-ness.
    863	 * Windows currently only uses 32 bits to save the clusters so we do
    864	 * the same as it is much faster on 32-bit CPUs.
    865	 */
    866	ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
    867	if ((u64)ll >= 1ULL << 32) {
    868		ntfs_error(vol->sb, "Cannot handle 64-bit clusters.  Sorry.");
    869		return false;
    870	}
    871	vol->nr_clusters = ll;
    872	ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
    873	/*
    874	 * On an architecture where unsigned long is 32-bits, we restrict the
    875	 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
    876	 * will hopefully optimize the whole check away.
    877	 */
    878	if (sizeof(unsigned long) < 8) {
    879		if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
    880			ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
    881					"large for this architecture.  "
    882					"Maximum supported is 2TiB.  Sorry.",
    883					(unsigned long long)ll >> (40 -
    884					vol->cluster_size_bits));
    885			return false;
    886		}
    887	}
    888	ll = sle64_to_cpu(b->mft_lcn);
    889	if (ll >= vol->nr_clusters) {
    890		ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
    891				"volume.  Weird.", (unsigned long long)ll,
    892				(unsigned long long)ll);
    893		return false;
    894	}
    895	vol->mft_lcn = ll;
    896	ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
    897	ll = sle64_to_cpu(b->mftmirr_lcn);
    898	if (ll >= vol->nr_clusters) {
    899		ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
    900				"of volume.  Weird.", (unsigned long long)ll,
    901				(unsigned long long)ll);
    902		return false;
    903	}
    904	vol->mftmirr_lcn = ll;
    905	ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
    906#ifdef NTFS_RW
    907	/*
    908	 * Work out the size of the mft mirror in number of mft records. If the
    909	 * cluster size is less than or equal to the size taken by four mft
    910	 * records, the mft mirror stores the first four mft records. If the
    911	 * cluster size is bigger than the size taken by four mft records, the
    912	 * mft mirror contains as many mft records as will fit into one
    913	 * cluster.
    914	 */
    915	if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
    916		vol->mftmirr_size = 4;
    917	else
    918		vol->mftmirr_size = vol->cluster_size >>
    919				vol->mft_record_size_bits;
    920	ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
    921#endif /* NTFS_RW */
    922	vol->serial_no = le64_to_cpu(b->volume_serial_number);
    923	ntfs_debug("vol->serial_no = 0x%llx",
    924			(unsigned long long)vol->serial_no);
    925	return true;
    926}
    927
    928/**
    929 * ntfs_setup_allocators - initialize the cluster and mft allocators
    930 * @vol:	volume structure for which to setup the allocators
    931 *
    932 * Setup the cluster (lcn) and mft allocators to the starting values.
    933 */
    934static void ntfs_setup_allocators(ntfs_volume *vol)
    935{
    936#ifdef NTFS_RW
    937	LCN mft_zone_size, mft_lcn;
    938#endif /* NTFS_RW */
    939
    940	ntfs_debug("vol->mft_zone_multiplier = 0x%x",
    941			vol->mft_zone_multiplier);
    942#ifdef NTFS_RW
    943	/* Determine the size of the MFT zone. */
    944	mft_zone_size = vol->nr_clusters;
    945	switch (vol->mft_zone_multiplier) {  /* % of volume size in clusters */
    946	case 4:
    947		mft_zone_size >>= 1;			/* 50%   */
    948		break;
    949	case 3:
    950		mft_zone_size = (mft_zone_size +
    951				(mft_zone_size >> 1)) >> 2;	/* 37.5% */
    952		break;
    953	case 2:
    954		mft_zone_size >>= 2;			/* 25%   */
    955		break;
    956	/* case 1: */
    957	default:
    958		mft_zone_size >>= 3;			/* 12.5% */
    959		break;
    960	}
    961	/* Setup the mft zone. */
    962	vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
    963	ntfs_debug("vol->mft_zone_pos = 0x%llx",
    964			(unsigned long long)vol->mft_zone_pos);
    965	/*
    966	 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
    967	 * source) and if the actual mft_lcn is in the expected place or even
    968	 * further to the front of the volume, extend the mft_zone to cover the
    969	 * beginning of the volume as well.  This is in order to protect the
    970	 * area reserved for the mft bitmap as well within the mft_zone itself.
    971	 * On non-standard volumes we do not protect it as the overhead would
    972	 * be higher than the speed increase we would get by doing it.
    973	 */
    974	mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
    975	if (mft_lcn * vol->cluster_size < 16 * 1024)
    976		mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
    977				vol->cluster_size;
    978	if (vol->mft_zone_start <= mft_lcn)
    979		vol->mft_zone_start = 0;
    980	ntfs_debug("vol->mft_zone_start = 0x%llx",
    981			(unsigned long long)vol->mft_zone_start);
    982	/*
    983	 * Need to cap the mft zone on non-standard volumes so that it does
    984	 * not point outside the boundaries of the volume.  We do this by
    985	 * halving the zone size until we are inside the volume.
    986	 */
    987	vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
    988	while (vol->mft_zone_end >= vol->nr_clusters) {
    989		mft_zone_size >>= 1;
    990		vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
    991	}
    992	ntfs_debug("vol->mft_zone_end = 0x%llx",
    993			(unsigned long long)vol->mft_zone_end);
    994	/*
    995	 * Set the current position within each data zone to the start of the
    996	 * respective zone.
    997	 */
    998	vol->data1_zone_pos = vol->mft_zone_end;
    999	ntfs_debug("vol->data1_zone_pos = 0x%llx",
   1000			(unsigned long long)vol->data1_zone_pos);
   1001	vol->data2_zone_pos = 0;
   1002	ntfs_debug("vol->data2_zone_pos = 0x%llx",
   1003			(unsigned long long)vol->data2_zone_pos);
   1004
   1005	/* Set the mft data allocation position to mft record 24. */
   1006	vol->mft_data_pos = 24;
   1007	ntfs_debug("vol->mft_data_pos = 0x%llx",
   1008			(unsigned long long)vol->mft_data_pos);
   1009#endif /* NTFS_RW */
   1010}
   1011
   1012#ifdef NTFS_RW
   1013
   1014/**
   1015 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
   1016 * @vol:	ntfs super block describing device whose mft mirror to load
   1017 *
   1018 * Return 'true' on success or 'false' on error.
   1019 */
   1020static bool load_and_init_mft_mirror(ntfs_volume *vol)
   1021{
   1022	struct inode *tmp_ino;
   1023	ntfs_inode *tmp_ni;
   1024
   1025	ntfs_debug("Entering.");
   1026	/* Get mft mirror inode. */
   1027	tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
   1028	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
   1029		if (!IS_ERR(tmp_ino))
   1030			iput(tmp_ino);
   1031		/* Caller will display error message. */
   1032		return false;
   1033	}
   1034	/*
   1035	 * Re-initialize some specifics about $MFTMirr's inode as
   1036	 * ntfs_read_inode() will have set up the default ones.
   1037	 */
   1038	/* Set uid and gid to root. */
   1039	tmp_ino->i_uid = GLOBAL_ROOT_UID;
   1040	tmp_ino->i_gid = GLOBAL_ROOT_GID;
   1041	/* Regular file.  No access for anyone. */
   1042	tmp_ino->i_mode = S_IFREG;
   1043	/* No VFS initiated operations allowed for $MFTMirr. */
   1044	tmp_ino->i_op = &ntfs_empty_inode_ops;
   1045	tmp_ino->i_fop = &ntfs_empty_file_ops;
   1046	/* Put in our special address space operations. */
   1047	tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
   1048	tmp_ni = NTFS_I(tmp_ino);
   1049	/* The $MFTMirr, like the $MFT is multi sector transfer protected. */
   1050	NInoSetMstProtected(tmp_ni);
   1051	NInoSetSparseDisabled(tmp_ni);
   1052	/*
   1053	 * Set up our little cheat allowing us to reuse the async read io
   1054	 * completion handler for directories.
   1055	 */
   1056	tmp_ni->itype.index.block_size = vol->mft_record_size;
   1057	tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
   1058	vol->mftmirr_ino = tmp_ino;
   1059	ntfs_debug("Done.");
   1060	return true;
   1061}
   1062
   1063/**
   1064 * check_mft_mirror - compare contents of the mft mirror with the mft
   1065 * @vol:	ntfs super block describing device whose mft mirror to check
   1066 *
   1067 * Return 'true' on success or 'false' on error.
   1068 *
   1069 * Note, this function also results in the mft mirror runlist being completely
   1070 * mapped into memory.  The mft mirror write code requires this and will BUG()
   1071 * should it find an unmapped runlist element.
   1072 */
   1073static bool check_mft_mirror(ntfs_volume *vol)
   1074{
   1075	struct super_block *sb = vol->sb;
   1076	ntfs_inode *mirr_ni;
   1077	struct page *mft_page, *mirr_page;
   1078	u8 *kmft, *kmirr;
   1079	runlist_element *rl, rl2[2];
   1080	pgoff_t index;
   1081	int mrecs_per_page, i;
   1082
   1083	ntfs_debug("Entering.");
   1084	/* Compare contents of $MFT and $MFTMirr. */
   1085	mrecs_per_page = PAGE_SIZE / vol->mft_record_size;
   1086	BUG_ON(!mrecs_per_page);
   1087	BUG_ON(!vol->mftmirr_size);
   1088	mft_page = mirr_page = NULL;
   1089	kmft = kmirr = NULL;
   1090	index = i = 0;
   1091	do {
   1092		u32 bytes;
   1093
   1094		/* Switch pages if necessary. */
   1095		if (!(i % mrecs_per_page)) {
   1096			if (index) {
   1097				ntfs_unmap_page(mft_page);
   1098				ntfs_unmap_page(mirr_page);
   1099			}
   1100			/* Get the $MFT page. */
   1101			mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
   1102					index);
   1103			if (IS_ERR(mft_page)) {
   1104				ntfs_error(sb, "Failed to read $MFT.");
   1105				return false;
   1106			}
   1107			kmft = page_address(mft_page);
   1108			/* Get the $MFTMirr page. */
   1109			mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
   1110					index);
   1111			if (IS_ERR(mirr_page)) {
   1112				ntfs_error(sb, "Failed to read $MFTMirr.");
   1113				goto mft_unmap_out;
   1114			}
   1115			kmirr = page_address(mirr_page);
   1116			++index;
   1117		}
   1118		/* Do not check the record if it is not in use. */
   1119		if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
   1120			/* Make sure the record is ok. */
   1121			if (ntfs_is_baad_recordp((le32*)kmft)) {
   1122				ntfs_error(sb, "Incomplete multi sector "
   1123						"transfer detected in mft "
   1124						"record %i.", i);
   1125mm_unmap_out:
   1126				ntfs_unmap_page(mirr_page);
   1127mft_unmap_out:
   1128				ntfs_unmap_page(mft_page);
   1129				return false;
   1130			}
   1131		}
   1132		/* Do not check the mirror record if it is not in use. */
   1133		if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
   1134			if (ntfs_is_baad_recordp((le32*)kmirr)) {
   1135				ntfs_error(sb, "Incomplete multi sector "
   1136						"transfer detected in mft "
   1137						"mirror record %i.", i);
   1138				goto mm_unmap_out;
   1139			}
   1140		}
   1141		/* Get the amount of data in the current record. */
   1142		bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
   1143		if (bytes < sizeof(MFT_RECORD_OLD) ||
   1144				bytes > vol->mft_record_size ||
   1145				ntfs_is_baad_recordp((le32*)kmft)) {
   1146			bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
   1147			if (bytes < sizeof(MFT_RECORD_OLD) ||
   1148					bytes > vol->mft_record_size ||
   1149					ntfs_is_baad_recordp((le32*)kmirr))
   1150				bytes = vol->mft_record_size;
   1151		}
   1152		/* Compare the two records. */
   1153		if (memcmp(kmft, kmirr, bytes)) {
   1154			ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
   1155					"match.  Run ntfsfix or chkdsk.", i);
   1156			goto mm_unmap_out;
   1157		}
   1158		kmft += vol->mft_record_size;
   1159		kmirr += vol->mft_record_size;
   1160	} while (++i < vol->mftmirr_size);
   1161	/* Release the last pages. */
   1162	ntfs_unmap_page(mft_page);
   1163	ntfs_unmap_page(mirr_page);
   1164
   1165	/* Construct the mft mirror runlist by hand. */
   1166	rl2[0].vcn = 0;
   1167	rl2[0].lcn = vol->mftmirr_lcn;
   1168	rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
   1169			vol->cluster_size - 1) / vol->cluster_size;
   1170	rl2[1].vcn = rl2[0].length;
   1171	rl2[1].lcn = LCN_ENOENT;
   1172	rl2[1].length = 0;
   1173	/*
   1174	 * Because we have just read all of the mft mirror, we know we have
   1175	 * mapped the full runlist for it.
   1176	 */
   1177	mirr_ni = NTFS_I(vol->mftmirr_ino);
   1178	down_read(&mirr_ni->runlist.lock);
   1179	rl = mirr_ni->runlist.rl;
   1180	/* Compare the two runlists.  They must be identical. */
   1181	i = 0;
   1182	do {
   1183		if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
   1184				rl2[i].length != rl[i].length) {
   1185			ntfs_error(sb, "$MFTMirr location mismatch.  "
   1186					"Run chkdsk.");
   1187			up_read(&mirr_ni->runlist.lock);
   1188			return false;
   1189		}
   1190	} while (rl2[i++].length);
   1191	up_read(&mirr_ni->runlist.lock);
   1192	ntfs_debug("Done.");
   1193	return true;
   1194}
   1195
   1196/**
   1197 * load_and_check_logfile - load and check the logfile inode for a volume
   1198 * @vol:	ntfs super block describing device whose logfile to load
   1199 *
   1200 * Return 'true' on success or 'false' on error.
   1201 */
   1202static bool load_and_check_logfile(ntfs_volume *vol,
   1203		RESTART_PAGE_HEADER **rp)
   1204{
   1205	struct inode *tmp_ino;
   1206
   1207	ntfs_debug("Entering.");
   1208	tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
   1209	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
   1210		if (!IS_ERR(tmp_ino))
   1211			iput(tmp_ino);
   1212		/* Caller will display error message. */
   1213		return false;
   1214	}
   1215	if (!ntfs_check_logfile(tmp_ino, rp)) {
   1216		iput(tmp_ino);
   1217		/* ntfs_check_logfile() will have displayed error output. */
   1218		return false;
   1219	}
   1220	NInoSetSparseDisabled(NTFS_I(tmp_ino));
   1221	vol->logfile_ino = tmp_ino;
   1222	ntfs_debug("Done.");
   1223	return true;
   1224}
   1225
   1226#define NTFS_HIBERFIL_HEADER_SIZE	4096
   1227
   1228/**
   1229 * check_windows_hibernation_status - check if Windows is suspended on a volume
   1230 * @vol:	ntfs super block of device to check
   1231 *
   1232 * Check if Windows is hibernated on the ntfs volume @vol.  This is done by
   1233 * looking for the file hiberfil.sys in the root directory of the volume.  If
   1234 * the file is not present Windows is definitely not suspended.
   1235 *
   1236 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
   1237 * definitely suspended (this volume is not the system volume).  Caveat:  on a
   1238 * system with many volumes it is possible that the < 4kiB check is bogus but
   1239 * for now this should do fine.
   1240 *
   1241 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
   1242 * hiberfil header (which is the first 4kiB).  If this begins with "hibr",
   1243 * Windows is definitely suspended.  If it is completely full of zeroes,
   1244 * Windows is definitely not hibernated.  Any other case is treated as if
   1245 * Windows is suspended.  This caters for the above mentioned caveat of a
   1246 * system with many volumes where no "hibr" magic would be present and there is
   1247 * no zero header.
   1248 *
   1249 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
   1250 * hibernated on the volume, and -errno on error.
   1251 */
   1252static int check_windows_hibernation_status(ntfs_volume *vol)
   1253{
   1254	MFT_REF mref;
   1255	struct inode *vi;
   1256	struct page *page;
   1257	u32 *kaddr, *kend;
   1258	ntfs_name *name = NULL;
   1259	int ret = 1;
   1260	static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
   1261			cpu_to_le16('i'), cpu_to_le16('b'),
   1262			cpu_to_le16('e'), cpu_to_le16('r'),
   1263			cpu_to_le16('f'), cpu_to_le16('i'),
   1264			cpu_to_le16('l'), cpu_to_le16('.'),
   1265			cpu_to_le16('s'), cpu_to_le16('y'),
   1266			cpu_to_le16('s'), 0 };
   1267
   1268	ntfs_debug("Entering.");
   1269	/*
   1270	 * Find the inode number for the hibernation file by looking up the
   1271	 * filename hiberfil.sys in the root directory.
   1272	 */
   1273	inode_lock(vol->root_ino);
   1274	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
   1275			&name);
   1276	inode_unlock(vol->root_ino);
   1277	if (IS_ERR_MREF(mref)) {
   1278		ret = MREF_ERR(mref);
   1279		/* If the file does not exist, Windows is not hibernated. */
   1280		if (ret == -ENOENT) {
   1281			ntfs_debug("hiberfil.sys not present.  Windows is not "
   1282					"hibernated on the volume.");
   1283			return 0;
   1284		}
   1285		/* A real error occurred. */
   1286		ntfs_error(vol->sb, "Failed to find inode number for "
   1287				"hiberfil.sys.");
   1288		return ret;
   1289	}
   1290	/* We do not care for the type of match that was found. */
   1291	kfree(name);
   1292	/* Get the inode. */
   1293	vi = ntfs_iget(vol->sb, MREF(mref));
   1294	if (IS_ERR(vi) || is_bad_inode(vi)) {
   1295		if (!IS_ERR(vi))
   1296			iput(vi);
   1297		ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
   1298		return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
   1299	}
   1300	if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
   1301		ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx).  "
   1302				"Windows is hibernated on the volume.  This "
   1303				"is not the system volume.", i_size_read(vi));
   1304		goto iput_out;
   1305	}
   1306	page = ntfs_map_page(vi->i_mapping, 0);
   1307	if (IS_ERR(page)) {
   1308		ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
   1309		ret = PTR_ERR(page);
   1310		goto iput_out;
   1311	}
   1312	kaddr = (u32*)page_address(page);
   1313	if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
   1314		ntfs_debug("Magic \"hibr\" found in hiberfil.sys.  Windows is "
   1315				"hibernated on the volume.  This is the "
   1316				"system volume.");
   1317		goto unm_iput_out;
   1318	}
   1319	kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
   1320	do {
   1321		if (unlikely(*kaddr)) {
   1322			ntfs_debug("hiberfil.sys is larger than 4kiB "
   1323					"(0x%llx), does not contain the "
   1324					"\"hibr\" magic, and does not have a "
   1325					"zero header.  Windows is hibernated "
   1326					"on the volume.  This is not the "
   1327					"system volume.", i_size_read(vi));
   1328			goto unm_iput_out;
   1329		}
   1330	} while (++kaddr < kend);
   1331	ntfs_debug("hiberfil.sys contains a zero header.  Windows is not "
   1332			"hibernated on the volume.  This is the system "
   1333			"volume.");
   1334	ret = 0;
   1335unm_iput_out:
   1336	ntfs_unmap_page(page);
   1337iput_out:
   1338	iput(vi);
   1339	return ret;
   1340}
   1341
   1342/**
   1343 * load_and_init_quota - load and setup the quota file for a volume if present
   1344 * @vol:	ntfs super block describing device whose quota file to load
   1345 *
   1346 * Return 'true' on success or 'false' on error.  If $Quota is not present, we
   1347 * leave vol->quota_ino as NULL and return success.
   1348 */
   1349static bool load_and_init_quota(ntfs_volume *vol)
   1350{
   1351	MFT_REF mref;
   1352	struct inode *tmp_ino;
   1353	ntfs_name *name = NULL;
   1354	static const ntfschar Quota[7] = { cpu_to_le16('$'),
   1355			cpu_to_le16('Q'), cpu_to_le16('u'),
   1356			cpu_to_le16('o'), cpu_to_le16('t'),
   1357			cpu_to_le16('a'), 0 };
   1358	static ntfschar Q[3] = { cpu_to_le16('$'),
   1359			cpu_to_le16('Q'), 0 };
   1360
   1361	ntfs_debug("Entering.");
   1362	/*
   1363	 * Find the inode number for the quota file by looking up the filename
   1364	 * $Quota in the extended system files directory $Extend.
   1365	 */
   1366	inode_lock(vol->extend_ino);
   1367	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
   1368			&name);
   1369	inode_unlock(vol->extend_ino);
   1370	if (IS_ERR_MREF(mref)) {
   1371		/*
   1372		 * If the file does not exist, quotas are disabled and have
   1373		 * never been enabled on this volume, just return success.
   1374		 */
   1375		if (MREF_ERR(mref) == -ENOENT) {
   1376			ntfs_debug("$Quota not present.  Volume does not have "
   1377					"quotas enabled.");
   1378			/*
   1379			 * No need to try to set quotas out of date if they are
   1380			 * not enabled.
   1381			 */
   1382			NVolSetQuotaOutOfDate(vol);
   1383			return true;
   1384		}
   1385		/* A real error occurred. */
   1386		ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
   1387		return false;
   1388	}
   1389	/* We do not care for the type of match that was found. */
   1390	kfree(name);
   1391	/* Get the inode. */
   1392	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
   1393	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
   1394		if (!IS_ERR(tmp_ino))
   1395			iput(tmp_ino);
   1396		ntfs_error(vol->sb, "Failed to load $Quota.");
   1397		return false;
   1398	}
   1399	vol->quota_ino = tmp_ino;
   1400	/* Get the $Q index allocation attribute. */
   1401	tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
   1402	if (IS_ERR(tmp_ino)) {
   1403		ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
   1404		return false;
   1405	}
   1406	vol->quota_q_ino = tmp_ino;
   1407	ntfs_debug("Done.");
   1408	return true;
   1409}
   1410
   1411/**
   1412 * load_and_init_usnjrnl - load and setup the transaction log if present
   1413 * @vol:	ntfs super block describing device whose usnjrnl file to load
   1414 *
   1415 * Return 'true' on success or 'false' on error.
   1416 *
   1417 * If $UsnJrnl is not present or in the process of being disabled, we set
   1418 * NVolUsnJrnlStamped() and return success.
   1419 *
   1420 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
   1421 * i.e. transaction logging has only just been enabled or the journal has been
   1422 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
   1423 * and return success.
   1424 */
   1425static bool load_and_init_usnjrnl(ntfs_volume *vol)
   1426{
   1427	MFT_REF mref;
   1428	struct inode *tmp_ino;
   1429	ntfs_inode *tmp_ni;
   1430	struct page *page;
   1431	ntfs_name *name = NULL;
   1432	USN_HEADER *uh;
   1433	static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
   1434			cpu_to_le16('U'), cpu_to_le16('s'),
   1435			cpu_to_le16('n'), cpu_to_le16('J'),
   1436			cpu_to_le16('r'), cpu_to_le16('n'),
   1437			cpu_to_le16('l'), 0 };
   1438	static ntfschar Max[5] = { cpu_to_le16('$'),
   1439			cpu_to_le16('M'), cpu_to_le16('a'),
   1440			cpu_to_le16('x'), 0 };
   1441	static ntfschar J[3] = { cpu_to_le16('$'),
   1442			cpu_to_le16('J'), 0 };
   1443
   1444	ntfs_debug("Entering.");
   1445	/*
   1446	 * Find the inode number for the transaction log file by looking up the
   1447	 * filename $UsnJrnl in the extended system files directory $Extend.
   1448	 */
   1449	inode_lock(vol->extend_ino);
   1450	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
   1451			&name);
   1452	inode_unlock(vol->extend_ino);
   1453	if (IS_ERR_MREF(mref)) {
   1454		/*
   1455		 * If the file does not exist, transaction logging is disabled,
   1456		 * just return success.
   1457		 */
   1458		if (MREF_ERR(mref) == -ENOENT) {
   1459			ntfs_debug("$UsnJrnl not present.  Volume does not "
   1460					"have transaction logging enabled.");
   1461not_enabled:
   1462			/*
   1463			 * No need to try to stamp the transaction log if
   1464			 * transaction logging is not enabled.
   1465			 */
   1466			NVolSetUsnJrnlStamped(vol);
   1467			return true;
   1468		}
   1469		/* A real error occurred. */
   1470		ntfs_error(vol->sb, "Failed to find inode number for "
   1471				"$UsnJrnl.");
   1472		return false;
   1473	}
   1474	/* We do not care for the type of match that was found. */
   1475	kfree(name);
   1476	/* Get the inode. */
   1477	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
   1478	if (IS_ERR(tmp_ino) || unlikely(is_bad_inode(tmp_ino))) {
   1479		if (!IS_ERR(tmp_ino))
   1480			iput(tmp_ino);
   1481		ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
   1482		return false;
   1483	}
   1484	vol->usnjrnl_ino = tmp_ino;
   1485	/*
   1486	 * If the transaction log is in the process of being deleted, we can
   1487	 * ignore it.
   1488	 */
   1489	if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
   1490		ntfs_debug("$UsnJrnl in the process of being disabled.  "
   1491				"Volume does not have transaction logging "
   1492				"enabled.");
   1493		goto not_enabled;
   1494	}
   1495	/* Get the $DATA/$Max attribute. */
   1496	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
   1497	if (IS_ERR(tmp_ino)) {
   1498		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
   1499				"attribute.");
   1500		return false;
   1501	}
   1502	vol->usnjrnl_max_ino = tmp_ino;
   1503	if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
   1504		ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
   1505				"attribute (size is 0x%llx but should be at "
   1506				"least 0x%zx bytes).", i_size_read(tmp_ino),
   1507				sizeof(USN_HEADER));
   1508		return false;
   1509	}
   1510	/* Get the $DATA/$J attribute. */
   1511	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
   1512	if (IS_ERR(tmp_ino)) {
   1513		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
   1514				"attribute.");
   1515		return false;
   1516	}
   1517	vol->usnjrnl_j_ino = tmp_ino;
   1518	/* Verify $J is non-resident and sparse. */
   1519	tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
   1520	if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
   1521		ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
   1522				"and/or not sparse.");
   1523		return false;
   1524	}
   1525	/* Read the USN_HEADER from $DATA/$Max. */
   1526	page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
   1527	if (IS_ERR(page)) {
   1528		ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
   1529				"attribute.");
   1530		return false;
   1531	}
   1532	uh = (USN_HEADER*)page_address(page);
   1533	/* Sanity check the $Max. */
   1534	if (unlikely(sle64_to_cpu(uh->allocation_delta) >
   1535			sle64_to_cpu(uh->maximum_size))) {
   1536		ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
   1537				"maximum size (0x%llx).  $UsnJrnl is corrupt.",
   1538				(long long)sle64_to_cpu(uh->allocation_delta),
   1539				(long long)sle64_to_cpu(uh->maximum_size));
   1540		ntfs_unmap_page(page);
   1541		return false;
   1542	}
   1543	/*
   1544	 * If the transaction log has been stamped and nothing has been written
   1545	 * to it since, we do not need to stamp it.
   1546	 */
   1547	if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
   1548			i_size_read(vol->usnjrnl_j_ino))) {
   1549		if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
   1550				i_size_read(vol->usnjrnl_j_ino))) {
   1551			ntfs_unmap_page(page);
   1552			ntfs_debug("$UsnJrnl is enabled but nothing has been "
   1553					"logged since it was last stamped.  "
   1554					"Treating this as if the volume does "
   1555					"not have transaction logging "
   1556					"enabled.");
   1557			goto not_enabled;
   1558		}
   1559		ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
   1560				"which is out of bounds (0x%llx).  $UsnJrnl "
   1561				"is corrupt.",
   1562				(long long)sle64_to_cpu(uh->lowest_valid_usn),
   1563				i_size_read(vol->usnjrnl_j_ino));
   1564		ntfs_unmap_page(page);
   1565		return false;
   1566	}
   1567	ntfs_unmap_page(page);
   1568	ntfs_debug("Done.");
   1569	return true;
   1570}
   1571
   1572/**
   1573 * load_and_init_attrdef - load the attribute definitions table for a volume
   1574 * @vol:	ntfs super block describing device whose attrdef to load
   1575 *
   1576 * Return 'true' on success or 'false' on error.
   1577 */
   1578static bool load_and_init_attrdef(ntfs_volume *vol)
   1579{
   1580	loff_t i_size;
   1581	struct super_block *sb = vol->sb;
   1582	struct inode *ino;
   1583	struct page *page;
   1584	pgoff_t index, max_index;
   1585	unsigned int size;
   1586
   1587	ntfs_debug("Entering.");
   1588	/* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
   1589	ino = ntfs_iget(sb, FILE_AttrDef);
   1590	if (IS_ERR(ino) || is_bad_inode(ino)) {
   1591		if (!IS_ERR(ino))
   1592			iput(ino);
   1593		goto failed;
   1594	}
   1595	NInoSetSparseDisabled(NTFS_I(ino));
   1596	/* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
   1597	i_size = i_size_read(ino);
   1598	if (i_size <= 0 || i_size > 0x7fffffff)
   1599		goto iput_failed;
   1600	vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
   1601	if (!vol->attrdef)
   1602		goto iput_failed;
   1603	index = 0;
   1604	max_index = i_size >> PAGE_SHIFT;
   1605	size = PAGE_SIZE;
   1606	while (index < max_index) {
   1607		/* Read the attrdef table and copy it into the linear buffer. */
   1608read_partial_attrdef_page:
   1609		page = ntfs_map_page(ino->i_mapping, index);
   1610		if (IS_ERR(page))
   1611			goto free_iput_failed;
   1612		memcpy((u8*)vol->attrdef + (index++ << PAGE_SHIFT),
   1613				page_address(page), size);
   1614		ntfs_unmap_page(page);
   1615	};
   1616	if (size == PAGE_SIZE) {
   1617		size = i_size & ~PAGE_MASK;
   1618		if (size)
   1619			goto read_partial_attrdef_page;
   1620	}
   1621	vol->attrdef_size = i_size;
   1622	ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
   1623	iput(ino);
   1624	return true;
   1625free_iput_failed:
   1626	ntfs_free(vol->attrdef);
   1627	vol->attrdef = NULL;
   1628iput_failed:
   1629	iput(ino);
   1630failed:
   1631	ntfs_error(sb, "Failed to initialize attribute definition table.");
   1632	return false;
   1633}
   1634
   1635#endif /* NTFS_RW */
   1636
   1637/**
   1638 * load_and_init_upcase - load the upcase table for an ntfs volume
   1639 * @vol:	ntfs super block describing device whose upcase to load
   1640 *
   1641 * Return 'true' on success or 'false' on error.
   1642 */
   1643static bool load_and_init_upcase(ntfs_volume *vol)
   1644{
   1645	loff_t i_size;
   1646	struct super_block *sb = vol->sb;
   1647	struct inode *ino;
   1648	struct page *page;
   1649	pgoff_t index, max_index;
   1650	unsigned int size;
   1651	int i, max;
   1652
   1653	ntfs_debug("Entering.");
   1654	/* Read upcase table and setup vol->upcase and vol->upcase_len. */
   1655	ino = ntfs_iget(sb, FILE_UpCase);
   1656	if (IS_ERR(ino) || is_bad_inode(ino)) {
   1657		if (!IS_ERR(ino))
   1658			iput(ino);
   1659		goto upcase_failed;
   1660	}
   1661	/*
   1662	 * The upcase size must not be above 64k Unicode characters, must not
   1663	 * be zero and must be a multiple of sizeof(ntfschar).
   1664	 */
   1665	i_size = i_size_read(ino);
   1666	if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
   1667			i_size > 64ULL * 1024 * sizeof(ntfschar))
   1668		goto iput_upcase_failed;
   1669	vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
   1670	if (!vol->upcase)
   1671		goto iput_upcase_failed;
   1672	index = 0;
   1673	max_index = i_size >> PAGE_SHIFT;
   1674	size = PAGE_SIZE;
   1675	while (index < max_index) {
   1676		/* Read the upcase table and copy it into the linear buffer. */
   1677read_partial_upcase_page:
   1678		page = ntfs_map_page(ino->i_mapping, index);
   1679		if (IS_ERR(page))
   1680			goto iput_upcase_failed;
   1681		memcpy((char*)vol->upcase + (index++ << PAGE_SHIFT),
   1682				page_address(page), size);
   1683		ntfs_unmap_page(page);
   1684	};
   1685	if (size == PAGE_SIZE) {
   1686		size = i_size & ~PAGE_MASK;
   1687		if (size)
   1688			goto read_partial_upcase_page;
   1689	}
   1690	vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
   1691	ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
   1692			i_size, 64 * 1024 * sizeof(ntfschar));
   1693	iput(ino);
   1694	mutex_lock(&ntfs_lock);
   1695	if (!default_upcase) {
   1696		ntfs_debug("Using volume specified $UpCase since default is "
   1697				"not present.");
   1698		mutex_unlock(&ntfs_lock);
   1699		return true;
   1700	}
   1701	max = default_upcase_len;
   1702	if (max > vol->upcase_len)
   1703		max = vol->upcase_len;
   1704	for (i = 0; i < max; i++)
   1705		if (vol->upcase[i] != default_upcase[i])
   1706			break;
   1707	if (i == max) {
   1708		ntfs_free(vol->upcase);
   1709		vol->upcase = default_upcase;
   1710		vol->upcase_len = max;
   1711		ntfs_nr_upcase_users++;
   1712		mutex_unlock(&ntfs_lock);
   1713		ntfs_debug("Volume specified $UpCase matches default. Using "
   1714				"default.");
   1715		return true;
   1716	}
   1717	mutex_unlock(&ntfs_lock);
   1718	ntfs_debug("Using volume specified $UpCase since it does not match "
   1719			"the default.");
   1720	return true;
   1721iput_upcase_failed:
   1722	iput(ino);
   1723	ntfs_free(vol->upcase);
   1724	vol->upcase = NULL;
   1725upcase_failed:
   1726	mutex_lock(&ntfs_lock);
   1727	if (default_upcase) {
   1728		vol->upcase = default_upcase;
   1729		vol->upcase_len = default_upcase_len;
   1730		ntfs_nr_upcase_users++;
   1731		mutex_unlock(&ntfs_lock);
   1732		ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
   1733				"default.");
   1734		return true;
   1735	}
   1736	mutex_unlock(&ntfs_lock);
   1737	ntfs_error(sb, "Failed to initialize upcase table.");
   1738	return false;
   1739}
   1740
   1741/*
   1742 * The lcn and mft bitmap inodes are NTFS-internal inodes with
   1743 * their own special locking rules:
   1744 */
   1745static struct lock_class_key
   1746	lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
   1747	mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
   1748
   1749/**
   1750 * load_system_files - open the system files using normal functions
   1751 * @vol:	ntfs super block describing device whose system files to load
   1752 *
   1753 * Open the system files with normal access functions and complete setting up
   1754 * the ntfs super block @vol.
   1755 *
   1756 * Return 'true' on success or 'false' on error.
   1757 */
   1758static bool load_system_files(ntfs_volume *vol)
   1759{
   1760	struct super_block *sb = vol->sb;
   1761	MFT_RECORD *m;
   1762	VOLUME_INFORMATION *vi;
   1763	ntfs_attr_search_ctx *ctx;
   1764#ifdef NTFS_RW
   1765	RESTART_PAGE_HEADER *rp;
   1766	int err;
   1767#endif /* NTFS_RW */
   1768
   1769	ntfs_debug("Entering.");
   1770#ifdef NTFS_RW
   1771	/* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
   1772	if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
   1773		static const char *es1 = "Failed to load $MFTMirr";
   1774		static const char *es2 = "$MFTMirr does not match $MFT";
   1775		static const char *es3 = ".  Run ntfsfix and/or chkdsk.";
   1776
   1777		/* If a read-write mount, convert it to a read-only mount. */
   1778		if (!sb_rdonly(sb)) {
   1779			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   1780					ON_ERRORS_CONTINUE))) {
   1781				ntfs_error(sb, "%s and neither on_errors="
   1782						"continue nor on_errors="
   1783						"remount-ro was specified%s",
   1784						!vol->mftmirr_ino ? es1 : es2,
   1785						es3);
   1786				goto iput_mirr_err_out;
   1787			}
   1788			sb->s_flags |= SB_RDONLY;
   1789			ntfs_error(sb, "%s.  Mounting read-only%s",
   1790					!vol->mftmirr_ino ? es1 : es2, es3);
   1791		} else
   1792			ntfs_warning(sb, "%s.  Will not be able to remount "
   1793					"read-write%s",
   1794					!vol->mftmirr_ino ? es1 : es2, es3);
   1795		/* This will prevent a read-write remount. */
   1796		NVolSetErrors(vol);
   1797	}
   1798#endif /* NTFS_RW */
   1799	/* Get mft bitmap attribute inode. */
   1800	vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
   1801	if (IS_ERR(vol->mftbmp_ino)) {
   1802		ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
   1803		goto iput_mirr_err_out;
   1804	}
   1805	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
   1806			   &mftbmp_runlist_lock_key);
   1807	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
   1808			   &mftbmp_mrec_lock_key);
   1809	/* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
   1810	if (!load_and_init_upcase(vol))
   1811		goto iput_mftbmp_err_out;
   1812#ifdef NTFS_RW
   1813	/*
   1814	 * Read attribute definitions table and setup @vol->attrdef and
   1815	 * @vol->attrdef_size.
   1816	 */
   1817	if (!load_and_init_attrdef(vol))
   1818		goto iput_upcase_err_out;
   1819#endif /* NTFS_RW */
   1820	/*
   1821	 * Get the cluster allocation bitmap inode and verify the size, no
   1822	 * need for any locking at this stage as we are already running
   1823	 * exclusively as we are mount in progress task.
   1824	 */
   1825	vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
   1826	if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
   1827		if (!IS_ERR(vol->lcnbmp_ino))
   1828			iput(vol->lcnbmp_ino);
   1829		goto bitmap_failed;
   1830	}
   1831	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
   1832			   &lcnbmp_runlist_lock_key);
   1833	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
   1834			   &lcnbmp_mrec_lock_key);
   1835
   1836	NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
   1837	if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
   1838		iput(vol->lcnbmp_ino);
   1839bitmap_failed:
   1840		ntfs_error(sb, "Failed to load $Bitmap.");
   1841		goto iput_attrdef_err_out;
   1842	}
   1843	/*
   1844	 * Get the volume inode and setup our cache of the volume flags and
   1845	 * version.
   1846	 */
   1847	vol->vol_ino = ntfs_iget(sb, FILE_Volume);
   1848	if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
   1849		if (!IS_ERR(vol->vol_ino))
   1850			iput(vol->vol_ino);
   1851volume_failed:
   1852		ntfs_error(sb, "Failed to load $Volume.");
   1853		goto iput_lcnbmp_err_out;
   1854	}
   1855	m = map_mft_record(NTFS_I(vol->vol_ino));
   1856	if (IS_ERR(m)) {
   1857iput_volume_failed:
   1858		iput(vol->vol_ino);
   1859		goto volume_failed;
   1860	}
   1861	if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
   1862		ntfs_error(sb, "Failed to get attribute search context.");
   1863		goto get_ctx_vol_failed;
   1864	}
   1865	if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
   1866			ctx) || ctx->attr->non_resident || ctx->attr->flags) {
   1867err_put_vol:
   1868		ntfs_attr_put_search_ctx(ctx);
   1869get_ctx_vol_failed:
   1870		unmap_mft_record(NTFS_I(vol->vol_ino));
   1871		goto iput_volume_failed;
   1872	}
   1873	vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
   1874			le16_to_cpu(ctx->attr->data.resident.value_offset));
   1875	/* Some bounds checks. */
   1876	if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
   1877			le32_to_cpu(ctx->attr->data.resident.value_length) >
   1878			(u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
   1879		goto err_put_vol;
   1880	/* Copy the volume flags and version to the ntfs_volume structure. */
   1881	vol->vol_flags = vi->flags;
   1882	vol->major_ver = vi->major_ver;
   1883	vol->minor_ver = vi->minor_ver;
   1884	ntfs_attr_put_search_ctx(ctx);
   1885	unmap_mft_record(NTFS_I(vol->vol_ino));
   1886	pr_info("volume version %i.%i.\n", vol->major_ver,
   1887			vol->minor_ver);
   1888	if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
   1889		ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
   1890				"volume version %i.%i (need at least version "
   1891				"3.0).", vol->major_ver, vol->minor_ver);
   1892		NVolClearSparseEnabled(vol);
   1893	}
   1894#ifdef NTFS_RW
   1895	/* Make sure that no unsupported volume flags are set. */
   1896	if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
   1897		static const char *es1a = "Volume is dirty";
   1898		static const char *es1b = "Volume has been modified by chkdsk";
   1899		static const char *es1c = "Volume has unsupported flags set";
   1900		static const char *es2a = ".  Run chkdsk and mount in Windows.";
   1901		static const char *es2b = ".  Mount in Windows.";
   1902		const char *es1, *es2;
   1903
   1904		es2 = es2a;
   1905		if (vol->vol_flags & VOLUME_IS_DIRTY)
   1906			es1 = es1a;
   1907		else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
   1908			es1 = es1b;
   1909			es2 = es2b;
   1910		} else {
   1911			es1 = es1c;
   1912			ntfs_warning(sb, "Unsupported volume flags 0x%x "
   1913					"encountered.",
   1914					(unsigned)le16_to_cpu(vol->vol_flags));
   1915		}
   1916		/* If a read-write mount, convert it to a read-only mount. */
   1917		if (!sb_rdonly(sb)) {
   1918			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   1919					ON_ERRORS_CONTINUE))) {
   1920				ntfs_error(sb, "%s and neither on_errors="
   1921						"continue nor on_errors="
   1922						"remount-ro was specified%s",
   1923						es1, es2);
   1924				goto iput_vol_err_out;
   1925			}
   1926			sb->s_flags |= SB_RDONLY;
   1927			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   1928		} else
   1929			ntfs_warning(sb, "%s.  Will not be able to remount "
   1930					"read-write%s", es1, es2);
   1931		/*
   1932		 * Do not set NVolErrors() because ntfs_remount() re-checks the
   1933		 * flags which we need to do in case any flags have changed.
   1934		 */
   1935	}
   1936	/*
   1937	 * Get the inode for the logfile, check it and determine if the volume
   1938	 * was shutdown cleanly.
   1939	 */
   1940	rp = NULL;
   1941	if (!load_and_check_logfile(vol, &rp) ||
   1942			!ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
   1943		static const char *es1a = "Failed to load $LogFile";
   1944		static const char *es1b = "$LogFile is not clean";
   1945		static const char *es2 = ".  Mount in Windows.";
   1946		const char *es1;
   1947
   1948		es1 = !vol->logfile_ino ? es1a : es1b;
   1949		/* If a read-write mount, convert it to a read-only mount. */
   1950		if (!sb_rdonly(sb)) {
   1951			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   1952					ON_ERRORS_CONTINUE))) {
   1953				ntfs_error(sb, "%s and neither on_errors="
   1954						"continue nor on_errors="
   1955						"remount-ro was specified%s",
   1956						es1, es2);
   1957				if (vol->logfile_ino) {
   1958					BUG_ON(!rp);
   1959					ntfs_free(rp);
   1960				}
   1961				goto iput_logfile_err_out;
   1962			}
   1963			sb->s_flags |= SB_RDONLY;
   1964			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   1965		} else
   1966			ntfs_warning(sb, "%s.  Will not be able to remount "
   1967					"read-write%s", es1, es2);
   1968		/* This will prevent a read-write remount. */
   1969		NVolSetErrors(vol);
   1970	}
   1971	ntfs_free(rp);
   1972#endif /* NTFS_RW */
   1973	/* Get the root directory inode so we can do path lookups. */
   1974	vol->root_ino = ntfs_iget(sb, FILE_root);
   1975	if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
   1976		if (!IS_ERR(vol->root_ino))
   1977			iput(vol->root_ino);
   1978		ntfs_error(sb, "Failed to load root directory.");
   1979		goto iput_logfile_err_out;
   1980	}
   1981#ifdef NTFS_RW
   1982	/*
   1983	 * Check if Windows is suspended to disk on the target volume.  If it
   1984	 * is hibernated, we must not write *anything* to the disk so set
   1985	 * NVolErrors() without setting the dirty volume flag and mount
   1986	 * read-only.  This will prevent read-write remounting and it will also
   1987	 * prevent all writes.
   1988	 */
   1989	err = check_windows_hibernation_status(vol);
   1990	if (unlikely(err)) {
   1991		static const char *es1a = "Failed to determine if Windows is "
   1992				"hibernated";
   1993		static const char *es1b = "Windows is hibernated";
   1994		static const char *es2 = ".  Run chkdsk.";
   1995		const char *es1;
   1996
   1997		es1 = err < 0 ? es1a : es1b;
   1998		/* If a read-write mount, convert it to a read-only mount. */
   1999		if (!sb_rdonly(sb)) {
   2000			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2001					ON_ERRORS_CONTINUE))) {
   2002				ntfs_error(sb, "%s and neither on_errors="
   2003						"continue nor on_errors="
   2004						"remount-ro was specified%s",
   2005						es1, es2);
   2006				goto iput_root_err_out;
   2007			}
   2008			sb->s_flags |= SB_RDONLY;
   2009			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2010		} else
   2011			ntfs_warning(sb, "%s.  Will not be able to remount "
   2012					"read-write%s", es1, es2);
   2013		/* This will prevent a read-write remount. */
   2014		NVolSetErrors(vol);
   2015	}
   2016	/* If (still) a read-write mount, mark the volume dirty. */
   2017	if (!sb_rdonly(sb) && ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
   2018		static const char *es1 = "Failed to set dirty bit in volume "
   2019				"information flags";
   2020		static const char *es2 = ".  Run chkdsk.";
   2021
   2022		/* Convert to a read-only mount. */
   2023		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2024				ON_ERRORS_CONTINUE))) {
   2025			ntfs_error(sb, "%s and neither on_errors=continue nor "
   2026					"on_errors=remount-ro was specified%s",
   2027					es1, es2);
   2028			goto iput_root_err_out;
   2029		}
   2030		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2031		sb->s_flags |= SB_RDONLY;
   2032		/*
   2033		 * Do not set NVolErrors() because ntfs_remount() might manage
   2034		 * to set the dirty flag in which case all would be well.
   2035		 */
   2036	}
   2037#if 0
   2038	// TODO: Enable this code once we start modifying anything that is
   2039	//	 different between NTFS 1.2 and 3.x...
   2040	/*
   2041	 * If (still) a read-write mount, set the NT4 compatibility flag on
   2042	 * newer NTFS version volumes.
   2043	 */
   2044	if (!(sb->s_flags & SB_RDONLY) && (vol->major_ver > 1) &&
   2045			ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
   2046		static const char *es1 = "Failed to set NT4 compatibility flag";
   2047		static const char *es2 = ".  Run chkdsk.";
   2048
   2049		/* Convert to a read-only mount. */
   2050		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2051				ON_ERRORS_CONTINUE))) {
   2052			ntfs_error(sb, "%s and neither on_errors=continue nor "
   2053					"on_errors=remount-ro was specified%s",
   2054					es1, es2);
   2055			goto iput_root_err_out;
   2056		}
   2057		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2058		sb->s_flags |= SB_RDONLY;
   2059		NVolSetErrors(vol);
   2060	}
   2061#endif
   2062	/* If (still) a read-write mount, empty the logfile. */
   2063	if (!sb_rdonly(sb) && !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 |= SB_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 $Quota";
   2105		static const char *es2 = ".  Run chkdsk.";
   2106
   2107		/* If a read-write mount, convert it to a read-only mount. */
   2108		if (!sb_rdonly(sb)) {
   2109			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2110					ON_ERRORS_CONTINUE))) {
   2111				ntfs_error(sb, "%s and neither on_errors="
   2112						"continue nor on_errors="
   2113						"remount-ro was specified%s",
   2114						es1, es2);
   2115				goto iput_quota_err_out;
   2116			}
   2117			sb->s_flags |= SB_RDONLY;
   2118			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2119		} else
   2120			ntfs_warning(sb, "%s.  Will not be able to remount "
   2121					"read-write%s", es1, es2);
   2122		/* This will prevent a read-write remount. */
   2123		NVolSetErrors(vol);
   2124	}
   2125	/* If (still) a read-write mount, mark the quotas out of date. */
   2126	if (!sb_rdonly(sb) && !ntfs_mark_quotas_out_of_date(vol)) {
   2127		static const char *es1 = "Failed to mark quotas out of date";
   2128		static const char *es2 = ".  Run chkdsk.";
   2129
   2130		/* Convert to a read-only mount. */
   2131		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2132				ON_ERRORS_CONTINUE))) {
   2133			ntfs_error(sb, "%s and neither on_errors=continue nor "
   2134					"on_errors=remount-ro was specified%s",
   2135					es1, es2);
   2136			goto iput_quota_err_out;
   2137		}
   2138		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2139		sb->s_flags |= SB_RDONLY;
   2140		NVolSetErrors(vol);
   2141	}
   2142	/*
   2143	 * Find the transaction log file ($UsnJrnl), load it if present, check
   2144	 * it, and set it up.
   2145	 */
   2146	if (!load_and_init_usnjrnl(vol)) {
   2147		static const char *es1 = "Failed to load $UsnJrnl";
   2148		static const char *es2 = ".  Run chkdsk.";
   2149
   2150		/* If a read-write mount, convert it to a read-only mount. */
   2151		if (!sb_rdonly(sb)) {
   2152			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2153					ON_ERRORS_CONTINUE))) {
   2154				ntfs_error(sb, "%s and neither on_errors="
   2155						"continue nor on_errors="
   2156						"remount-ro was specified%s",
   2157						es1, es2);
   2158				goto iput_usnjrnl_err_out;
   2159			}
   2160			sb->s_flags |= SB_RDONLY;
   2161			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2162		} else
   2163			ntfs_warning(sb, "%s.  Will not be able to remount "
   2164					"read-write%s", es1, es2);
   2165		/* This will prevent a read-write remount. */
   2166		NVolSetErrors(vol);
   2167	}
   2168	/* If (still) a read-write mount, stamp the transaction log. */
   2169	if (!sb_rdonly(sb) && !ntfs_stamp_usnjrnl(vol)) {
   2170		static const char *es1 = "Failed to stamp transaction log "
   2171				"($UsnJrnl)";
   2172		static const char *es2 = ".  Run chkdsk.";
   2173
   2174		/* Convert to a read-only mount. */
   2175		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
   2176				ON_ERRORS_CONTINUE))) {
   2177			ntfs_error(sb, "%s and neither on_errors=continue nor "
   2178					"on_errors=remount-ro was specified%s",
   2179					es1, es2);
   2180			goto iput_usnjrnl_err_out;
   2181		}
   2182		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
   2183		sb->s_flags |= SB_RDONLY;
   2184		NVolSetErrors(vol);
   2185	}
   2186#endif /* NTFS_RW */
   2187	return true;
   2188#ifdef NTFS_RW
   2189iput_usnjrnl_err_out:
   2190	iput(vol->usnjrnl_j_ino);
   2191	iput(vol->usnjrnl_max_ino);
   2192	iput(vol->usnjrnl_ino);
   2193iput_quota_err_out:
   2194	iput(vol->quota_q_ino);
   2195	iput(vol->quota_ino);
   2196	iput(vol->extend_ino);
   2197#endif /* NTFS_RW */
   2198iput_sec_err_out:
   2199	iput(vol->secure_ino);
   2200iput_root_err_out:
   2201	iput(vol->root_ino);
   2202iput_logfile_err_out:
   2203#ifdef NTFS_RW
   2204	iput(vol->logfile_ino);
   2205iput_vol_err_out:
   2206#endif /* NTFS_RW */
   2207	iput(vol->vol_ino);
   2208iput_lcnbmp_err_out:
   2209	iput(vol->lcnbmp_ino);
   2210iput_attrdef_err_out:
   2211	vol->attrdef_size = 0;
   2212	if (vol->attrdef) {
   2213		ntfs_free(vol->attrdef);
   2214		vol->attrdef = NULL;
   2215	}
   2216#ifdef NTFS_RW
   2217iput_upcase_err_out:
   2218#endif /* NTFS_RW */
   2219	vol->upcase_len = 0;
   2220	mutex_lock(&ntfs_lock);
   2221	if (vol->upcase == default_upcase) {
   2222		ntfs_nr_upcase_users--;
   2223		vol->upcase = NULL;
   2224	}
   2225	mutex_unlock(&ntfs_lock);
   2226	if (vol->upcase) {
   2227		ntfs_free(vol->upcase);
   2228		vol->upcase = NULL;
   2229	}
   2230iput_mftbmp_err_out:
   2231	iput(vol->mftbmp_ino);
   2232iput_mirr_err_out:
   2233#ifdef NTFS_RW
   2234	iput(vol->mftmirr_ino);
   2235#endif /* NTFS_RW */
   2236	return false;
   2237}
   2238
   2239/**
   2240 * ntfs_put_super - called by the vfs to unmount a volume
   2241 * @sb:		vfs superblock of volume to unmount
   2242 *
   2243 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
   2244 * the volume is being unmounted (umount system call has been invoked) and it
   2245 * releases all inodes and memory belonging to the NTFS specific part of the
   2246 * super block.
   2247 */
   2248static void ntfs_put_super(struct super_block *sb)
   2249{
   2250	ntfs_volume *vol = NTFS_SB(sb);
   2251
   2252	ntfs_debug("Entering.");
   2253
   2254#ifdef NTFS_RW
   2255	/*
   2256	 * Commit all inodes while they are still open in case some of them
   2257	 * cause others to be dirtied.
   2258	 */
   2259	ntfs_commit_inode(vol->vol_ino);
   2260
   2261	/* NTFS 3.0+ specific. */
   2262	if (vol->major_ver >= 3) {
   2263		if (vol->usnjrnl_j_ino)
   2264			ntfs_commit_inode(vol->usnjrnl_j_ino);
   2265		if (vol->usnjrnl_max_ino)
   2266			ntfs_commit_inode(vol->usnjrnl_max_ino);
   2267		if (vol->usnjrnl_ino)
   2268			ntfs_commit_inode(vol->usnjrnl_ino);
   2269		if (vol->quota_q_ino)
   2270			ntfs_commit_inode(vol->quota_q_ino);
   2271		if (vol->quota_ino)
   2272			ntfs_commit_inode(vol->quota_ino);
   2273		if (vol->extend_ino)
   2274			ntfs_commit_inode(vol->extend_ino);
   2275		if (vol->secure_ino)
   2276			ntfs_commit_inode(vol->secure_ino);
   2277	}
   2278
   2279	ntfs_commit_inode(vol->root_ino);
   2280
   2281	down_write(&vol->lcnbmp_lock);
   2282	ntfs_commit_inode(vol->lcnbmp_ino);
   2283	up_write(&vol->lcnbmp_lock);
   2284
   2285	down_write(&vol->mftbmp_lock);
   2286	ntfs_commit_inode(vol->mftbmp_ino);
   2287	up_write(&vol->mftbmp_lock);
   2288
   2289	if (vol->logfile_ino)
   2290		ntfs_commit_inode(vol->logfile_ino);
   2291
   2292	if (vol->mftmirr_ino)
   2293		ntfs_commit_inode(vol->mftmirr_ino);
   2294	ntfs_commit_inode(vol->mft_ino);
   2295
   2296	/*
   2297	 * If a read-write mount and no volume errors have occurred, mark the
   2298	 * volume clean.  Also, re-commit all affected inodes.
   2299	 */
   2300	if (!sb_rdonly(sb)) {
   2301		if (!NVolErrors(vol)) {
   2302			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
   2303				ntfs_warning(sb, "Failed to clear dirty bit "
   2304						"in volume information "
   2305						"flags.  Run chkdsk.");
   2306			ntfs_commit_inode(vol->vol_ino);
   2307			ntfs_commit_inode(vol->root_ino);
   2308			if (vol->mftmirr_ino)
   2309				ntfs_commit_inode(vol->mftmirr_ino);
   2310			ntfs_commit_inode(vol->mft_ino);
   2311		} else {
   2312			ntfs_warning(sb, "Volume has errors.  Leaving volume "
   2313					"marked dirty.  Run chkdsk.");
   2314		}
   2315	}
   2316#endif /* NTFS_RW */
   2317
   2318	iput(vol->vol_ino);
   2319	vol->vol_ino = NULL;
   2320
   2321	/* NTFS 3.0+ specific clean up. */
   2322	if (vol->major_ver >= 3) {
   2323#ifdef NTFS_RW
   2324		if (vol->usnjrnl_j_ino) {
   2325			iput(vol->usnjrnl_j_ino);
   2326			vol->usnjrnl_j_ino = NULL;
   2327		}
   2328		if (vol->usnjrnl_max_ino) {
   2329			iput(vol->usnjrnl_max_ino);
   2330			vol->usnjrnl_max_ino = NULL;
   2331		}
   2332		if (vol->usnjrnl_ino) {
   2333			iput(vol->usnjrnl_ino);
   2334			vol->usnjrnl_ino = NULL;
   2335		}
   2336		if (vol->quota_q_ino) {
   2337			iput(vol->quota_q_ino);
   2338			vol->quota_q_ino = NULL;
   2339		}
   2340		if (vol->quota_ino) {
   2341			iput(vol->quota_ino);
   2342			vol->quota_ino = NULL;
   2343		}
   2344#endif /* NTFS_RW */
   2345		if (vol->extend_ino) {
   2346			iput(vol->extend_ino);
   2347			vol->extend_ino = NULL;
   2348		}
   2349		if (vol->secure_ino) {
   2350			iput(vol->secure_ino);
   2351			vol->secure_ino = NULL;
   2352		}
   2353	}
   2354
   2355	iput(vol->root_ino);
   2356	vol->root_ino = NULL;
   2357
   2358	down_write(&vol->lcnbmp_lock);
   2359	iput(vol->lcnbmp_ino);
   2360	vol->lcnbmp_ino = NULL;
   2361	up_write(&vol->lcnbmp_lock);
   2362
   2363	down_write(&vol->mftbmp_lock);
   2364	iput(vol->mftbmp_ino);
   2365	vol->mftbmp_ino = NULL;
   2366	up_write(&vol->mftbmp_lock);
   2367
   2368#ifdef NTFS_RW
   2369	if (vol->logfile_ino) {
   2370		iput(vol->logfile_ino);
   2371		vol->logfile_ino = NULL;
   2372	}
   2373	if (vol->mftmirr_ino) {
   2374		/* Re-commit the mft mirror and mft just in case. */
   2375		ntfs_commit_inode(vol->mftmirr_ino);
   2376		ntfs_commit_inode(vol->mft_ino);
   2377		iput(vol->mftmirr_ino);
   2378		vol->mftmirr_ino = NULL;
   2379	}
   2380	/*
   2381	 * We should have no dirty inodes left, due to
   2382	 * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
   2383	 * the underlying mft records are written out and cleaned.
   2384	 */
   2385	ntfs_commit_inode(vol->mft_ino);
   2386	write_inode_now(vol->mft_ino, 1);
   2387#endif /* NTFS_RW */
   2388
   2389	iput(vol->mft_ino);
   2390	vol->mft_ino = NULL;
   2391
   2392	/* Throw away the table of attribute definitions. */
   2393	vol->attrdef_size = 0;
   2394	if (vol->attrdef) {
   2395		ntfs_free(vol->attrdef);
   2396		vol->attrdef = NULL;
   2397	}
   2398	vol->upcase_len = 0;
   2399	/*
   2400	 * Destroy the global default upcase table if necessary.  Also decrease
   2401	 * the number of upcase users if we are a user.
   2402	 */
   2403	mutex_lock(&ntfs_lock);
   2404	if (vol->upcase == default_upcase) {
   2405		ntfs_nr_upcase_users--;
   2406		vol->upcase = NULL;
   2407	}
   2408	if (!ntfs_nr_upcase_users && default_upcase) {
   2409		ntfs_free(default_upcase);
   2410		default_upcase = NULL;
   2411	}
   2412	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
   2413		free_compression_buffers();
   2414	mutex_unlock(&ntfs_lock);
   2415	if (vol->upcase) {
   2416		ntfs_free(vol->upcase);
   2417		vol->upcase = NULL;
   2418	}
   2419
   2420	unload_nls(vol->nls_map);
   2421
   2422	sb->s_fs_info = NULL;
   2423	kfree(vol);
   2424}
   2425
   2426/**
   2427 * get_nr_free_clusters - return the number of free clusters on a volume
   2428 * @vol:	ntfs volume for which to obtain free cluster count
   2429 *
   2430 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
   2431 * actually calculate the number of clusters in use instead because this
   2432 * allows us to not care about partial pages as these will be just zero filled
   2433 * and hence not be counted as allocated clusters.
   2434 *
   2435 * The only particularity is that clusters beyond the end of the logical ntfs
   2436 * volume will be marked as allocated to prevent errors which means we have to
   2437 * discount those at the end. This is important as the cluster bitmap always
   2438 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
   2439 * the logical volume and marked in use when they are not as they do not exist.
   2440 *
   2441 * If any pages cannot be read we assume all clusters in the erroring pages are
   2442 * in use. This means we return an underestimate on errors which is better than
   2443 * an overestimate.
   2444 */
   2445static s64 get_nr_free_clusters(ntfs_volume *vol)
   2446{
   2447	s64 nr_free = vol->nr_clusters;
   2448	struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
   2449	struct page *page;
   2450	pgoff_t index, max_index;
   2451
   2452	ntfs_debug("Entering.");
   2453	/* Serialize accesses to the cluster bitmap. */
   2454	down_read(&vol->lcnbmp_lock);
   2455	/*
   2456	 * Convert the number of bits into bytes rounded up, then convert into
   2457	 * multiples of PAGE_SIZE, rounding up so that if we have one
   2458	 * full and one partial page max_index = 2.
   2459	 */
   2460	max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_SIZE - 1) >>
   2461			PAGE_SHIFT;
   2462	/* Use multiples of 4 bytes, thus max_size is PAGE_SIZE / 4. */
   2463	ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
   2464			max_index, PAGE_SIZE / 4);
   2465	for (index = 0; index < max_index; index++) {
   2466		unsigned long *kaddr;
   2467
   2468		/*
   2469		 * Read the page from page cache, getting it from backing store
   2470		 * if necessary, and increment the use count.
   2471		 */
   2472		page = read_mapping_page(mapping, index, NULL);
   2473		/* Ignore pages which errored synchronously. */
   2474		if (IS_ERR(page)) {
   2475			ntfs_debug("read_mapping_page() error. Skipping "
   2476					"page (index 0x%lx).", index);
   2477			nr_free -= PAGE_SIZE * 8;
   2478			continue;
   2479		}
   2480		kaddr = kmap_atomic(page);
   2481		/*
   2482		 * Subtract the number of set bits. If this
   2483		 * is the last page and it is partial we don't really care as
   2484		 * it just means we do a little extra work but it won't affect
   2485		 * the result as all out of range bytes are set to zero by
   2486		 * ntfs_readpage().
   2487		 */
   2488		nr_free -= bitmap_weight(kaddr,
   2489					PAGE_SIZE * BITS_PER_BYTE);
   2490		kunmap_atomic(kaddr);
   2491		put_page(page);
   2492	}
   2493	ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
   2494	/*
   2495	 * Fixup for eventual bits outside logical ntfs volume (see function
   2496	 * description above).
   2497	 */
   2498	if (vol->nr_clusters & 63)
   2499		nr_free += 64 - (vol->nr_clusters & 63);
   2500	up_read(&vol->lcnbmp_lock);
   2501	/* If errors occurred we may well have gone below zero, fix this. */
   2502	if (nr_free < 0)
   2503		nr_free = 0;
   2504	ntfs_debug("Exiting.");
   2505	return nr_free;
   2506}
   2507
   2508/**
   2509 * __get_nr_free_mft_records - return the number of free inodes on a volume
   2510 * @vol:	ntfs volume for which to obtain free inode count
   2511 * @nr_free:	number of mft records in filesystem
   2512 * @max_index:	maximum number of pages containing set bits
   2513 *
   2514 * Calculate the number of free mft records (inodes) on the mounted NTFS
   2515 * volume @vol. We actually calculate the number of mft records in use instead
   2516 * because this allows us to not care about partial pages as these will be just
   2517 * zero filled and hence not be counted as allocated mft record.
   2518 *
   2519 * If any pages cannot be read we assume all mft records in the erroring pages
   2520 * are in use. This means we return an underestimate on errors which is better
   2521 * than an overestimate.
   2522 *
   2523 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
   2524 */
   2525static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
   2526		s64 nr_free, const pgoff_t max_index)
   2527{
   2528	struct address_space *mapping = vol->mftbmp_ino->i_mapping;
   2529	struct page *page;
   2530	pgoff_t index;
   2531
   2532	ntfs_debug("Entering.");
   2533	/* Use multiples of 4 bytes, thus max_size is PAGE_SIZE / 4. */
   2534	ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
   2535			"0x%lx.", max_index, PAGE_SIZE / 4);
   2536	for (index = 0; index < max_index; index++) {
   2537		unsigned long *kaddr;
   2538
   2539		/*
   2540		 * Read the page from page cache, getting it from backing store
   2541		 * if necessary, and increment the use count.
   2542		 */
   2543		page = read_mapping_page(mapping, index, NULL);
   2544		/* Ignore pages which errored synchronously. */
   2545		if (IS_ERR(page)) {
   2546			ntfs_debug("read_mapping_page() error. Skipping "
   2547					"page (index 0x%lx).", index);
   2548			nr_free -= PAGE_SIZE * 8;
   2549			continue;
   2550		}
   2551		kaddr = kmap_atomic(page);
   2552		/*
   2553		 * Subtract the number of set bits. If this
   2554		 * is the last page and it is partial we don't really care as
   2555		 * it just means we do a little extra work but it won't affect
   2556		 * the result as all out of range bytes are set to zero by
   2557		 * ntfs_readpage().
   2558		 */
   2559		nr_free -= bitmap_weight(kaddr,
   2560					PAGE_SIZE * BITS_PER_BYTE);
   2561		kunmap_atomic(kaddr);
   2562		put_page(page);
   2563	}
   2564	ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
   2565			index - 1);
   2566	/* If errors occurred we may well have gone below zero, fix this. */
   2567	if (nr_free < 0)
   2568		nr_free = 0;
   2569	ntfs_debug("Exiting.");
   2570	return nr_free;
   2571}
   2572
   2573/**
   2574 * ntfs_statfs - return information about mounted NTFS volume
   2575 * @dentry:	dentry from mounted volume
   2576 * @sfs:	statfs structure in which to return the information
   2577 *
   2578 * Return information about the mounted NTFS volume @dentry in the statfs structure
   2579 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
   2580 * called). We interpret the values to be correct of the moment in time at
   2581 * which we are called. Most values are variable otherwise and this isn't just
   2582 * the free values but the totals as well. For example we can increase the
   2583 * total number of file nodes if we run out and we can keep doing this until
   2584 * there is no more space on the volume left at all.
   2585 *
   2586 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
   2587 * ustat system calls.
   2588 *
   2589 * Return 0 on success or -errno on error.
   2590 */
   2591static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
   2592{
   2593	struct super_block *sb = dentry->d_sb;
   2594	s64 size;
   2595	ntfs_volume *vol = NTFS_SB(sb);
   2596	ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
   2597	pgoff_t max_index;
   2598	unsigned long flags;
   2599
   2600	ntfs_debug("Entering.");
   2601	/* Type of filesystem. */
   2602	sfs->f_type   = NTFS_SB_MAGIC;
   2603	/* Optimal transfer block size. */
   2604	sfs->f_bsize  = PAGE_SIZE;
   2605	/*
   2606	 * Total data blocks in filesystem in units of f_bsize and since
   2607	 * inodes are also stored in data blocs ($MFT is a file) this is just
   2608	 * the total clusters.
   2609	 */
   2610	sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
   2611				PAGE_SHIFT;
   2612	/* Free data blocks in filesystem in units of f_bsize. */
   2613	size	      = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
   2614				PAGE_SHIFT;
   2615	if (size < 0LL)
   2616		size = 0LL;
   2617	/* Free blocks avail to non-superuser, same as above on NTFS. */
   2618	sfs->f_bavail = sfs->f_bfree = size;
   2619	/* Serialize accesses to the inode bitmap. */
   2620	down_read(&vol->mftbmp_lock);
   2621	read_lock_irqsave(&mft_ni->size_lock, flags);
   2622	size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
   2623	/*
   2624	 * Convert the maximum number of set bits into bytes rounded up, then
   2625	 * convert into multiples of PAGE_SIZE, rounding up so that if we
   2626	 * have one full and one partial page max_index = 2.
   2627	 */
   2628	max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
   2629			+ 7) >> 3) + PAGE_SIZE - 1) >> PAGE_SHIFT;
   2630	read_unlock_irqrestore(&mft_ni->size_lock, flags);
   2631	/* Number of inodes in filesystem (at this point in time). */
   2632	sfs->f_files = size;
   2633	/* Free inodes in fs (based on current total count). */
   2634	sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
   2635	up_read(&vol->mftbmp_lock);
   2636	/*
   2637	 * File system id. This is extremely *nix flavour dependent and even
   2638	 * within Linux itself all fs do their own thing. I interpret this to
   2639	 * mean a unique id associated with the mounted fs and not the id
   2640	 * associated with the filesystem driver, the latter is already given
   2641	 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
   2642	 * volume serial number splitting it into two 32-bit parts. We enter
   2643	 * the least significant 32-bits in f_fsid[0] and the most significant
   2644	 * 32-bits in f_fsid[1].
   2645	 */
   2646	sfs->f_fsid = u64_to_fsid(vol->serial_no);
   2647	/* Maximum length of filenames. */
   2648	sfs->f_namelen	   = NTFS_MAX_NAME_LEN;
   2649	return 0;
   2650}
   2651
   2652#ifdef NTFS_RW
   2653static int ntfs_write_inode(struct inode *vi, struct writeback_control *wbc)
   2654{
   2655	return __ntfs_write_inode(vi, wbc->sync_mode == WB_SYNC_ALL);
   2656}
   2657#endif
   2658
   2659/**
   2660 * The complete super operations.
   2661 */
   2662static const struct super_operations ntfs_sops = {
   2663	.alloc_inode	= ntfs_alloc_big_inode,	  /* VFS: Allocate new inode. */
   2664	.free_inode	= ntfs_free_big_inode, /* VFS: Deallocate inode. */
   2665#ifdef NTFS_RW
   2666	.write_inode	= ntfs_write_inode,	/* VFS: Write dirty inode to
   2667						   disk. */
   2668#endif /* NTFS_RW */
   2669	.put_super	= ntfs_put_super,	/* Syscall: umount. */
   2670	.statfs		= ntfs_statfs,		/* Syscall: statfs */
   2671	.remount_fs	= ntfs_remount,		/* Syscall: mount -o remount. */
   2672	.evict_inode	= ntfs_evict_big_inode,	/* VFS: Called when an inode is
   2673						   removed from memory. */
   2674	.show_options	= ntfs_show_options,	/* Show mount options in
   2675						   proc. */
   2676};
   2677
   2678/**
   2679 * ntfs_fill_super - mount an ntfs filesystem
   2680 * @sb:		super block of ntfs filesystem to mount
   2681 * @opt:	string containing the mount options
   2682 * @silent:	silence error output
   2683 *
   2684 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
   2685 * with the mount otions in @data with the NTFS filesystem.
   2686 *
   2687 * If @silent is true, remain silent even if errors are detected. This is used
   2688 * during bootup, when the kernel tries to mount the root filesystem with all
   2689 * registered filesystems one after the other until one succeeds. This implies
   2690 * that all filesystems except the correct one will quite correctly and
   2691 * expectedly return an error, but nobody wants to see error messages when in
   2692 * fact this is what is supposed to happen.
   2693 *
   2694 * NOTE: @sb->s_flags contains the mount options flags.
   2695 */
   2696static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
   2697{
   2698	ntfs_volume *vol;
   2699	struct buffer_head *bh;
   2700	struct inode *tmp_ino;
   2701	int blocksize, result;
   2702
   2703	/*
   2704	 * We do a pretty difficult piece of bootstrap by reading the
   2705	 * MFT (and other metadata) from disk into memory. We'll only
   2706	 * release this metadata during umount, so the locking patterns
   2707	 * observed during bootstrap do not count. So turn off the
   2708	 * observation of locking patterns (strictly for this context
   2709	 * only) while mounting NTFS. [The validator is still active
   2710	 * otherwise, even for this context: it will for example record
   2711	 * lock class registrations.]
   2712	 */
   2713	lockdep_off();
   2714	ntfs_debug("Entering.");
   2715#ifndef NTFS_RW
   2716	sb->s_flags |= SB_RDONLY;
   2717#endif /* ! NTFS_RW */
   2718	/* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
   2719	sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
   2720	vol = NTFS_SB(sb);
   2721	if (!vol) {
   2722		if (!silent)
   2723			ntfs_error(sb, "Allocation of NTFS volume structure "
   2724					"failed. Aborting mount...");
   2725		lockdep_on();
   2726		return -ENOMEM;
   2727	}
   2728	/* Initialize ntfs_volume structure. */
   2729	*vol = (ntfs_volume) {
   2730		.sb = sb,
   2731		/*
   2732		 * Default is group and other don't have any access to files or
   2733		 * directories while owner has full access. Further, files by
   2734		 * default are not executable but directories are of course
   2735		 * browseable.
   2736		 */
   2737		.fmask = 0177,
   2738		.dmask = 0077,
   2739	};
   2740	init_rwsem(&vol->mftbmp_lock);
   2741	init_rwsem(&vol->lcnbmp_lock);
   2742
   2743	/* By default, enable sparse support. */
   2744	NVolSetSparseEnabled(vol);
   2745
   2746	/* Important to get the mount options dealt with now. */
   2747	if (!parse_options(vol, (char*)opt))
   2748		goto err_out_now;
   2749
   2750	/* We support sector sizes up to the PAGE_SIZE. */
   2751	if (bdev_logical_block_size(sb->s_bdev) > PAGE_SIZE) {
   2752		if (!silent)
   2753			ntfs_error(sb, "Device has unsupported sector size "
   2754					"(%i).  The maximum supported sector "
   2755					"size on this architecture is %lu "
   2756					"bytes.",
   2757					bdev_logical_block_size(sb->s_bdev),
   2758					PAGE_SIZE);
   2759		goto err_out_now;
   2760	}
   2761	/*
   2762	 * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
   2763	 * sector size, whichever is bigger.
   2764	 */
   2765	blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
   2766	if (blocksize < NTFS_BLOCK_SIZE) {
   2767		if (!silent)
   2768			ntfs_error(sb, "Unable to set device block size.");
   2769		goto err_out_now;
   2770	}
   2771	BUG_ON(blocksize != sb->s_blocksize);
   2772	ntfs_debug("Set device block size to %i bytes (block size bits %i).",
   2773			blocksize, sb->s_blocksize_bits);
   2774	/* Determine the size of the device in units of block_size bytes. */
   2775	vol->nr_blocks = sb_bdev_nr_blocks(sb);
   2776	if (!vol->nr_blocks) {
   2777		if (!silent)
   2778			ntfs_error(sb, "Unable to determine device size.");
   2779		goto err_out_now;
   2780	}
   2781	/* Read the boot sector and return unlocked buffer head to it. */
   2782	if (!(bh = read_ntfs_boot_sector(sb, silent))) {
   2783		if (!silent)
   2784			ntfs_error(sb, "Not an NTFS volume.");
   2785		goto err_out_now;
   2786	}
   2787	/*
   2788	 * Extract the data from the boot sector and setup the ntfs volume
   2789	 * using it.
   2790	 */
   2791	result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
   2792	brelse(bh);
   2793	if (!result) {
   2794		if (!silent)
   2795			ntfs_error(sb, "Unsupported NTFS filesystem.");
   2796		goto err_out_now;
   2797	}
   2798	/*
   2799	 * If the boot sector indicates a sector size bigger than the current
   2800	 * device block size, switch the device block size to the sector size.
   2801	 * TODO: It may be possible to support this case even when the set
   2802	 * below fails, we would just be breaking up the i/o for each sector
   2803	 * into multiple blocks for i/o purposes but otherwise it should just
   2804	 * work.  However it is safer to leave disabled until someone hits this
   2805	 * error message and then we can get them to try it without the setting
   2806	 * so we know for sure that it works.
   2807	 */
   2808	if (vol->sector_size > blocksize) {
   2809		blocksize = sb_set_blocksize(sb, vol->sector_size);
   2810		if (blocksize != vol->sector_size) {
   2811			if (!silent)
   2812				ntfs_error(sb, "Unable to set device block "
   2813						"size to sector size (%i).",
   2814						vol->sector_size);
   2815			goto err_out_now;
   2816		}
   2817		BUG_ON(blocksize != sb->s_blocksize);
   2818		vol->nr_blocks = sb_bdev_nr_blocks(sb);
   2819		ntfs_debug("Changed device block size to %i bytes (block size "
   2820				"bits %i) to match volume sector size.",
   2821				blocksize, sb->s_blocksize_bits);
   2822	}
   2823	/* Initialize the cluster and mft allocators. */
   2824	ntfs_setup_allocators(vol);
   2825	/* Setup remaining fields in the super block. */
   2826	sb->s_magic = NTFS_SB_MAGIC;
   2827	/*
   2828	 * Ntfs allows 63 bits for the file size, i.e. correct would be:
   2829	 *	sb->s_maxbytes = ~0ULL >> 1;
   2830	 * But the kernel uses a long as the page cache page index which on
   2831	 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
   2832	 * defined to the maximum the page cache page index can cope with
   2833	 * without overflowing the index or to 2^63 - 1, whichever is smaller.
   2834	 */
   2835	sb->s_maxbytes = MAX_LFS_FILESIZE;
   2836	/* Ntfs measures time in 100ns intervals. */
   2837	sb->s_time_gran = 100;
   2838	/*
   2839	 * Now load the metadata required for the page cache and our address
   2840	 * space operations to function. We do this by setting up a specialised
   2841	 * read_inode method and then just calling the normal iget() to obtain
   2842	 * the inode for $MFT which is sufficient to allow our normal inode
   2843	 * operations and associated address space operations to function.
   2844	 */
   2845	sb->s_op = &ntfs_sops;
   2846	tmp_ino = new_inode(sb);
   2847	if (!tmp_ino) {
   2848		if (!silent)
   2849			ntfs_error(sb, "Failed to load essential metadata.");
   2850		goto err_out_now;
   2851	}
   2852	tmp_ino->i_ino = FILE_MFT;
   2853	insert_inode_hash(tmp_ino);
   2854	if (ntfs_read_inode_mount(tmp_ino) < 0) {
   2855		if (!silent)
   2856			ntfs_error(sb, "Failed to load essential metadata.");
   2857		goto iput_tmp_ino_err_out_now;
   2858	}
   2859	mutex_lock(&ntfs_lock);
   2860	/*
   2861	 * The current mount is a compression user if the cluster size is
   2862	 * less than or equal 4kiB.
   2863	 */
   2864	if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
   2865		result = allocate_compression_buffers();
   2866		if (result) {
   2867			ntfs_error(NULL, "Failed to allocate buffers "
   2868					"for compression engine.");
   2869			ntfs_nr_compression_users--;
   2870			mutex_unlock(&ntfs_lock);
   2871			goto iput_tmp_ino_err_out_now;
   2872		}
   2873	}
   2874	/*
   2875	 * Generate the global default upcase table if necessary.  Also
   2876	 * temporarily increment the number of upcase users to avoid race
   2877	 * conditions with concurrent (u)mounts.
   2878	 */
   2879	if (!default_upcase)
   2880		default_upcase = generate_default_upcase();
   2881	ntfs_nr_upcase_users++;
   2882	mutex_unlock(&ntfs_lock);
   2883	/*
   2884	 * From now on, ignore @silent parameter. If we fail below this line,
   2885	 * it will be due to a corrupt fs or a system error, so we report it.
   2886	 */
   2887	/*
   2888	 * Open the system files with normal access functions and complete
   2889	 * setting up the ntfs super block.
   2890	 */
   2891	if (!load_system_files(vol)) {
   2892		ntfs_error(sb, "Failed to load system files.");
   2893		goto unl_upcase_iput_tmp_ino_err_out_now;
   2894	}
   2895
   2896	/* We grab a reference, simulating an ntfs_iget(). */
   2897	ihold(vol->root_ino);
   2898	if ((sb->s_root = d_make_root(vol->root_ino))) {
   2899		ntfs_debug("Exiting, status successful.");
   2900		/* Release the default upcase if it has no users. */
   2901		mutex_lock(&ntfs_lock);
   2902		if (!--ntfs_nr_upcase_users && default_upcase) {
   2903			ntfs_free(default_upcase);
   2904			default_upcase = NULL;
   2905		}
   2906		mutex_unlock(&ntfs_lock);
   2907		sb->s_export_op = &ntfs_export_ops;
   2908		lockdep_on();
   2909		return 0;
   2910	}
   2911	ntfs_error(sb, "Failed to allocate root directory.");
   2912	/* Clean up after the successful load_system_files() call from above. */
   2913	// TODO: Use ntfs_put_super() instead of repeating all this code...
   2914	// FIXME: Should mark the volume clean as the error is most likely
   2915	// 	  -ENOMEM.
   2916	iput(vol->vol_ino);
   2917	vol->vol_ino = NULL;
   2918	/* NTFS 3.0+ specific clean up. */
   2919	if (vol->major_ver >= 3) {
   2920#ifdef NTFS_RW
   2921		if (vol->usnjrnl_j_ino) {
   2922			iput(vol->usnjrnl_j_ino);
   2923			vol->usnjrnl_j_ino = NULL;
   2924		}
   2925		if (vol->usnjrnl_max_ino) {
   2926			iput(vol->usnjrnl_max_ino);
   2927			vol->usnjrnl_max_ino = NULL;
   2928		}
   2929		if (vol->usnjrnl_ino) {
   2930			iput(vol->usnjrnl_ino);
   2931			vol->usnjrnl_ino = NULL;
   2932		}
   2933		if (vol->quota_q_ino) {
   2934			iput(vol->quota_q_ino);
   2935			vol->quota_q_ino = NULL;
   2936		}
   2937		if (vol->quota_ino) {
   2938			iput(vol->quota_ino);
   2939			vol->quota_ino = NULL;
   2940		}
   2941#endif /* NTFS_RW */
   2942		if (vol->extend_ino) {
   2943			iput(vol->extend_ino);
   2944			vol->extend_ino = NULL;
   2945		}
   2946		if (vol->secure_ino) {
   2947			iput(vol->secure_ino);
   2948			vol->secure_ino = NULL;
   2949		}
   2950	}
   2951	iput(vol->root_ino);
   2952	vol->root_ino = NULL;
   2953	iput(vol->lcnbmp_ino);
   2954	vol->lcnbmp_ino = NULL;
   2955	iput(vol->mftbmp_ino);
   2956	vol->mftbmp_ino = NULL;
   2957#ifdef NTFS_RW
   2958	if (vol->logfile_ino) {
   2959		iput(vol->logfile_ino);
   2960		vol->logfile_ino = NULL;
   2961	}
   2962	if (vol->mftmirr_ino) {
   2963		iput(vol->mftmirr_ino);
   2964		vol->mftmirr_ino = NULL;
   2965	}
   2966#endif /* NTFS_RW */
   2967	/* Throw away the table of attribute definitions. */
   2968	vol->attrdef_size = 0;
   2969	if (vol->attrdef) {
   2970		ntfs_free(vol->attrdef);
   2971		vol->attrdef = NULL;
   2972	}
   2973	vol->upcase_len = 0;
   2974	mutex_lock(&ntfs_lock);
   2975	if (vol->upcase == default_upcase) {
   2976		ntfs_nr_upcase_users--;
   2977		vol->upcase = NULL;
   2978	}
   2979	mutex_unlock(&ntfs_lock);
   2980	if (vol->upcase) {
   2981		ntfs_free(vol->upcase);
   2982		vol->upcase = NULL;
   2983	}
   2984	if (vol->nls_map) {
   2985		unload_nls(vol->nls_map);
   2986		vol->nls_map = NULL;
   2987	}
   2988	/* Error exit code path. */
   2989unl_upcase_iput_tmp_ino_err_out_now:
   2990	/*
   2991	 * Decrease the number of upcase users and destroy the global default
   2992	 * upcase table if necessary.
   2993	 */
   2994	mutex_lock(&ntfs_lock);
   2995	if (!--ntfs_nr_upcase_users && default_upcase) {
   2996		ntfs_free(default_upcase);
   2997		default_upcase = NULL;
   2998	}
   2999	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
   3000		free_compression_buffers();
   3001	mutex_unlock(&ntfs_lock);
   3002iput_tmp_ino_err_out_now:
   3003	iput(tmp_ino);
   3004	if (vol->mft_ino && vol->mft_ino != tmp_ino)
   3005		iput(vol->mft_ino);
   3006	vol->mft_ino = NULL;
   3007	/* Errors at this stage are irrelevant. */
   3008err_out_now:
   3009	sb->s_fs_info = NULL;
   3010	kfree(vol);
   3011	ntfs_debug("Failed, returning -EINVAL.");
   3012	lockdep_on();
   3013	return -EINVAL;
   3014}
   3015
   3016/*
   3017 * This is a slab cache to optimize allocations and deallocations of Unicode
   3018 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
   3019 * (255) Unicode characters + a terminating NULL Unicode character.
   3020 */
   3021struct kmem_cache *ntfs_name_cache;
   3022
   3023/* Slab caches for efficient allocation/deallocation of inodes. */
   3024struct kmem_cache *ntfs_inode_cache;
   3025struct kmem_cache *ntfs_big_inode_cache;
   3026
   3027/* Init once constructor for the inode slab cache. */
   3028static void ntfs_big_inode_init_once(void *foo)
   3029{
   3030	ntfs_inode *ni = (ntfs_inode *)foo;
   3031
   3032	inode_init_once(VFS_I(ni));
   3033}
   3034
   3035/*
   3036 * Slab caches to optimize allocations and deallocations of attribute search
   3037 * contexts and index contexts, respectively.
   3038 */
   3039struct kmem_cache *ntfs_attr_ctx_cache;
   3040struct kmem_cache *ntfs_index_ctx_cache;
   3041
   3042/* Driver wide mutex. */
   3043DEFINE_MUTEX(ntfs_lock);
   3044
   3045static struct dentry *ntfs_mount(struct file_system_type *fs_type,
   3046	int flags, const char *dev_name, void *data)
   3047{
   3048	return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
   3049}
   3050
   3051static struct file_system_type ntfs_fs_type = {
   3052	.owner		= THIS_MODULE,
   3053	.name		= "ntfs",
   3054	.mount		= ntfs_mount,
   3055	.kill_sb	= kill_block_super,
   3056	.fs_flags	= FS_REQUIRES_DEV,
   3057};
   3058MODULE_ALIAS_FS("ntfs");
   3059
   3060/* Stable names for the slab caches. */
   3061static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
   3062static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
   3063static const char ntfs_name_cache_name[] = "ntfs_name_cache";
   3064static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
   3065static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
   3066
   3067static int __init init_ntfs_fs(void)
   3068{
   3069	int err = 0;
   3070
   3071	/* This may be ugly but it results in pretty output so who cares. (-8 */
   3072	pr_info("driver " NTFS_VERSION " [Flags: R/"
   3073#ifdef NTFS_RW
   3074			"W"
   3075#else
   3076			"O"
   3077#endif
   3078#ifdef DEBUG
   3079			" DEBUG"
   3080#endif
   3081#ifdef MODULE
   3082			" MODULE"
   3083#endif
   3084			"].\n");
   3085
   3086	ntfs_debug("Debug messages are enabled.");
   3087
   3088	ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
   3089			sizeof(ntfs_index_context), 0 /* offset */,
   3090			SLAB_HWCACHE_ALIGN, NULL /* ctor */);
   3091	if (!ntfs_index_ctx_cache) {
   3092		pr_crit("Failed to create %s!\n", ntfs_index_ctx_cache_name);
   3093		goto ictx_err_out;
   3094	}
   3095	ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
   3096			sizeof(ntfs_attr_search_ctx), 0 /* offset */,
   3097			SLAB_HWCACHE_ALIGN, NULL /* ctor */);
   3098	if (!ntfs_attr_ctx_cache) {
   3099		pr_crit("NTFS: Failed to create %s!\n",
   3100			ntfs_attr_ctx_cache_name);
   3101		goto actx_err_out;
   3102	}
   3103
   3104	ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
   3105			(NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
   3106			SLAB_HWCACHE_ALIGN, NULL);
   3107	if (!ntfs_name_cache) {
   3108		pr_crit("Failed to create %s!\n", ntfs_name_cache_name);
   3109		goto name_err_out;
   3110	}
   3111
   3112	ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
   3113			sizeof(ntfs_inode), 0,
   3114			SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
   3115	if (!ntfs_inode_cache) {
   3116		pr_crit("Failed to create %s!\n", ntfs_inode_cache_name);
   3117		goto inode_err_out;
   3118	}
   3119
   3120	ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
   3121			sizeof(big_ntfs_inode), 0,
   3122			SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD|
   3123			SLAB_ACCOUNT, ntfs_big_inode_init_once);
   3124	if (!ntfs_big_inode_cache) {
   3125		pr_crit("Failed to create %s!\n", ntfs_big_inode_cache_name);
   3126		goto big_inode_err_out;
   3127	}
   3128
   3129	/* Register the ntfs sysctls. */
   3130	err = ntfs_sysctl(1);
   3131	if (err) {
   3132		pr_crit("Failed to register NTFS sysctls!\n");
   3133		goto sysctl_err_out;
   3134	}
   3135
   3136	err = register_filesystem(&ntfs_fs_type);
   3137	if (!err) {
   3138		ntfs_debug("NTFS driver registered successfully.");
   3139		return 0; /* Success! */
   3140	}
   3141	pr_crit("Failed to register NTFS filesystem driver!\n");
   3142
   3143	/* Unregister the ntfs sysctls. */
   3144	ntfs_sysctl(0);
   3145sysctl_err_out:
   3146	kmem_cache_destroy(ntfs_big_inode_cache);
   3147big_inode_err_out:
   3148	kmem_cache_destroy(ntfs_inode_cache);
   3149inode_err_out:
   3150	kmem_cache_destroy(ntfs_name_cache);
   3151name_err_out:
   3152	kmem_cache_destroy(ntfs_attr_ctx_cache);
   3153actx_err_out:
   3154	kmem_cache_destroy(ntfs_index_ctx_cache);
   3155ictx_err_out:
   3156	if (!err) {
   3157		pr_crit("Aborting NTFS filesystem driver registration...\n");
   3158		err = -ENOMEM;
   3159	}
   3160	return err;
   3161}
   3162
   3163static void __exit exit_ntfs_fs(void)
   3164{
   3165	ntfs_debug("Unregistering NTFS driver.");
   3166
   3167	unregister_filesystem(&ntfs_fs_type);
   3168
   3169	/*
   3170	 * Make sure all delayed rcu free inodes are flushed before we
   3171	 * destroy cache.
   3172	 */
   3173	rcu_barrier();
   3174	kmem_cache_destroy(ntfs_big_inode_cache);
   3175	kmem_cache_destroy(ntfs_inode_cache);
   3176	kmem_cache_destroy(ntfs_name_cache);
   3177	kmem_cache_destroy(ntfs_attr_ctx_cache);
   3178	kmem_cache_destroy(ntfs_index_ctx_cache);
   3179	/* Unregister the ntfs sysctls. */
   3180	ntfs_sysctl(0);
   3181}
   3182
   3183MODULE_AUTHOR("Anton Altaparmakov <anton@tuxera.com>");
   3184MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2014 Anton Altaparmakov and Tuxera Inc.");
   3185MODULE_VERSION(NTFS_VERSION);
   3186MODULE_LICENSE("GPL");
   3187#ifdef DEBUG
   3188module_param(debug_msgs, bint, 0);
   3189MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
   3190#endif
   3191
   3192module_init(init_ntfs_fs)
   3193module_exit(exit_ntfs_fs)