1/*
2 * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
3 *
4 * Copyright (c) 2001-2007 Anton Altaparmakov
5 * Copyright (c) 2001,2002 Richard Russon
6 *
7 * This program/include file is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License as published
9 * by the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program/include file is distributed in the hope that it will be
13 * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program (in the main directory of the Linux-NTFS
19 * distribution in the file COPYING); if not, write to the Free Software
20 * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
21 */
22
23#include <linux/stddef.h>
24#include <linux/init.h>
25#include <linux/slab.h>
26#include <linux/string.h>
27#include <linux/spinlock.h>
28#include <linux/blkdev.h>	/* For bdev_hardsect_size(). */
29#include <linux/backing-dev.h>
30#include <linux/buffer_head.h>
31#include <linux/vfs.h>
32#include <linux/moduleparam.h>
33#include <linux/smp_lock.h>
34
35#include "sysctl.h"
36#include "logfile.h"
37#include "quota.h"
38#include "usnjrnl.h"
39#include "dir.h"
40#include "debug.h"
41#include "index.h"
42#include "aops.h"
43#include "layout.h"
44#include "malloc.h"
45#include "ntfs.h"
46
47/* Number of mounted filesystems which have compression enabled. */
48static unsigned long ntfs_nr_compression_users;
49
50/* A global default upcase table and a corresponding reference count. */
51static ntfschar *default_upcase = NULL;
52static unsigned long ntfs_nr_upcase_users = 0;
53
54/* Error constants/strings used in inode.c::ntfs_show_options(). */
55typedef enum {
56	/* One of these must be present, default is ON_ERRORS_CONTINUE. */
57	ON_ERRORS_PANIC			= 0x01,
58	ON_ERRORS_REMOUNT_RO		= 0x02,
59	ON_ERRORS_CONTINUE		= 0x04,
60	/* Optional, can be combined with any of the above. */
61	ON_ERRORS_RECOVER		= 0x10,
62} ON_ERRORS_ACTIONS;
63
64const option_t on_errors_arr[] = {
65	{ ON_ERRORS_PANIC,	"panic" },
66	{ ON_ERRORS_REMOUNT_RO,	"remount-ro", },
67	{ ON_ERRORS_CONTINUE,	"continue", },
68	{ ON_ERRORS_RECOVER,	"recover" },
69	{ 0,			NULL }
70};
71
72/**
73 * simple_getbool -
74 *
75 * Copied from old ntfs driver (which copied from vfat driver).
76 */
77static int simple_getbool(char *s, bool *setval)
78{
79	if (s) {
80		if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
81			*setval = true;
82		else if (!strcmp(s, "0") || !strcmp(s, "no") ||
83							!strcmp(s, "false"))
84			*setval = false;
85		else
86			return 0;
87	} else
88		*setval = true;
89	return 1;
90}
91
92/**
93 * parse_options - parse the (re)mount options
94 * @vol:	ntfs volume
95 * @opt:	string containing the (re)mount options
96 *
97 * Parse the recognized options in @opt for the ntfs volume described by @vol.
98 */
99static bool parse_options(ntfs_volume *vol, char *opt)
100{
101	char *p, *v, *ov;
102	static char *utf8 = "utf8";
103	int errors = 0, sloppy = 0;
104	uid_t uid = (uid_t)-1;
105	gid_t gid = (gid_t)-1;
106	mode_t fmask = (mode_t)-1, dmask = (mode_t)-1;
107	int mft_zone_multiplier = -1, on_errors = -1;
108	int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
109	struct nls_table *nls_map = NULL, *old_nls;
110
111	/* I am lazy... (-8 */
112#define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value)	\
113	if (!strcmp(p, option)) {					\
114		if (!v || !*v)						\
115			variable = default_value;			\
116		else {							\
117			variable = simple_strtoul(ov = v, &v, 0);	\
118			if (*v)						\
119				goto needs_val;				\
120		}							\
121	}
122#define NTFS_GETOPT(option, variable)					\
123	if (!strcmp(p, option)) {					\
124		if (!v || !*v)						\
125			goto needs_arg;					\
126		variable = simple_strtoul(ov = v, &v, 0);		\
127		if (*v)							\
128			goto needs_val;					\
129	}
130#define NTFS_GETOPT_OCTAL(option, variable)				\
131	if (!strcmp(p, option)) {					\
132		if (!v || !*v)						\
133			goto needs_arg;					\
134		variable = simple_strtoul(ov = v, &v, 8);		\
135		if (*v)							\
136			goto needs_val;					\
137	}
138#define NTFS_GETOPT_BOOL(option, variable)				\
139	if (!strcmp(p, option)) {					\
140		bool val;						\
141		if (!simple_getbool(v, &val))				\
142			goto needs_bool;				\
143		variable = val;						\
144	}
145#define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array)		\
146	if (!strcmp(p, option)) {					\
147		int _i;							\
148		if (!v || !*v)						\
149			goto needs_arg;					\
150		ov = v;							\
151		if (variable == -1)					\
152			variable = 0;					\
153		for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
154			if (!strcmp(opt_array[_i].str, v)) {		\
155				variable |= opt_array[_i].val;		\
156				break;					\
157			}						\
158		if (!opt_array[_i].str || !*opt_array[_i].str)		\
159			goto needs_val;					\
160	}
161	if (!opt || !*opt)
162		goto no_mount_options;
163	ntfs_debug("Entering with mount options string: %s", opt);
164	while ((p = strsep(&opt, ","))) {
165		if ((v = strchr(p, '=')))
166			*v++ = 0;
167		NTFS_GETOPT("uid", uid)
168		else NTFS_GETOPT("gid", gid)
169		else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
170		else NTFS_GETOPT_OCTAL("fmask", fmask)
171		else NTFS_GETOPT_OCTAL("dmask", dmask)
172		else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
173		else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
174		else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
175		else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
176		else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
177		else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
178				on_errors_arr)
179		else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
180			ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
181					p);
182		else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
183			if (!strcmp(p, "iocharset"))
184				ntfs_warning(vol->sb, "Option iocharset is "
185						"deprecated. Please use "
186						"option nls=<charsetname> in "
187						"the future.");
188			if (!v || !*v)
189				goto needs_arg;
190use_utf8:
191			old_nls = nls_map;
192			nls_map = load_nls(v);
193			if (!nls_map) {
194				if (!old_nls) {
195					ntfs_error(vol->sb, "NLS character set "
196							"%s not found.", v);
197					return false;
198				}
199				ntfs_error(vol->sb, "NLS character set %s not "
200						"found. Using previous one %s.",
201						v, old_nls->charset);
202				nls_map = old_nls;
203			} else /* nls_map */ {
204				if (old_nls)
205					unload_nls(old_nls);
206			}
207		} else if (!strcmp(p, "utf8")) {
208			bool val = false;
209			ntfs_warning(vol->sb, "Option utf8 is no longer "
210				   "supported, using option nls=utf8. Please "
211				   "use option nls=utf8 in the future and "
212				   "make sure utf8 is compiled either as a "
213				   "module or into the kernel.");
214			if (!v || !*v)
215				val = true;
216			else if (!simple_getbool(v, &val))
217				goto needs_bool;
218			if (val) {
219				v = utf8;
220				goto use_utf8;
221			}
222		} else {
223			ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
224			if (errors < INT_MAX)
225				errors++;
226		}
227#undef NTFS_GETOPT_OPTIONS_ARRAY
228#undef NTFS_GETOPT_BOOL
229#undef NTFS_GETOPT
230#undef NTFS_GETOPT_WITH_DEFAULT
231	}
232no_mount_options:
233	if (errors && !sloppy)
234		return false;
235	if (sloppy)
236		ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
237				"unrecognized mount option(s) and continuing.");
238	/* Keep this first! */
239	if (on_errors != -1) {
240		if (!on_errors) {
241			ntfs_error(vol->sb, "Invalid errors option argument "
242					"or bug in options parser.");
243			return false;
244		}
245	}
246	if (nls_map) {
247		if (vol->nls_map && vol->nls_map != nls_map) {
248			ntfs_error(vol->sb, "Cannot change NLS character set "
249					"on remount.");
250			return false;
251		} /* else (!vol->nls_map) */
252		ntfs_debug("Using NLS character set %s.", nls_map->charset);
253		vol->nls_map = nls_map;
254	} else /* (!nls_map) */ {
255		if (!vol->nls_map) {
256			vol->nls_map = load_nls_default();
257			if (!vol->nls_map) {
258				ntfs_error(vol->sb, "Failed to load default "
259						"NLS character set.");
260				return false;
261			}
262			ntfs_debug("Using default NLS character set (%s).",
263					vol->nls_map->charset);
264		}
265	}
266	if (mft_zone_multiplier != -1) {
267		if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
268				mft_zone_multiplier) {
269			ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
270					"on remount.");
271			return false;
272		}
273		if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
274			ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
275					"Using default value, i.e. 1.");
276			mft_zone_multiplier = 1;
277		}
278		vol->mft_zone_multiplier = mft_zone_multiplier;
279	}
280	if (!vol->mft_zone_multiplier)
281		vol->mft_zone_multiplier = 1;
282	if (on_errors != -1)
283		vol->on_errors = on_errors;
284	if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
285		vol->on_errors |= ON_ERRORS_CONTINUE;
286	if (uid != (uid_t)-1)
287		vol->uid = uid;
288	if (gid != (gid_t)-1)
289		vol->gid = gid;
290	if (fmask != (mode_t)-1)
291		vol->fmask = fmask;
292	if (dmask != (mode_t)-1)
293		vol->dmask = dmask;
294	if (show_sys_files != -1) {
295		if (show_sys_files)
296			NVolSetShowSystemFiles(vol);
297		else
298			NVolClearShowSystemFiles(vol);
299	}
300	if (case_sensitive != -1) {
301		if (case_sensitive)
302			NVolSetCaseSensitive(vol);
303		else
304			NVolClearCaseSensitive(vol);
305	}
306	if (disable_sparse != -1) {
307		if (disable_sparse)
308			NVolClearSparseEnabled(vol);
309		else {
310			if (!NVolSparseEnabled(vol) &&
311					vol->major_ver && vol->major_ver < 3)
312				ntfs_warning(vol->sb, "Not enabling sparse "
313						"support due to NTFS volume "
314						"version %i.%i (need at least "
315						"version 3.0).", vol->major_ver,
316						vol->minor_ver);
317			else
318				NVolSetSparseEnabled(vol);
319		}
320	}
321	return true;
322needs_arg:
323	ntfs_error(vol->sb, "The %s option requires an argument.", p);
324	return false;
325needs_bool:
326	ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
327	return false;
328needs_val:
329	ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
330	return false;
331}
332
333#ifdef NTFS_RW
334
335/**
336 * ntfs_write_volume_flags - write new flags to the volume information flags
337 * @vol:	ntfs volume on which to modify the flags
338 * @flags:	new flags value for the volume information flags
339 *
340 * Internal function.  You probably want to use ntfs_{set,clear}_volume_flags()
341 * instead (see below).
342 *
343 * Replace the volume information flags on the volume @vol with the value
344 * supplied in @flags.  Note, this overwrites the volume information flags, so
345 * make sure to combine the flags you want to modify with the old flags and use
346 * the result when calling ntfs_write_volume_flags().
347 *
348 * Return 0 on success and -errno on error.
349 */
350static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
351{
352	ntfs_inode *ni = NTFS_I(vol->vol_ino);
353	MFT_RECORD *m;
354	VOLUME_INFORMATION *vi;
355	ntfs_attr_search_ctx *ctx;
356	int err;
357
358	ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
359			le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
360	if (vol->vol_flags == flags)
361		goto done;
362	BUG_ON(!ni);
363	m = map_mft_record(ni);
364	if (IS_ERR(m)) {
365		err = PTR_ERR(m);
366		goto err_out;
367	}
368	ctx = ntfs_attr_get_search_ctx(ni, m);
369	if (!ctx) {
370		err = -ENOMEM;
371		goto put_unm_err_out;
372	}
373	err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
374			ctx);
375	if (err)
376		goto put_unm_err_out;
377	vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
378			le16_to_cpu(ctx->attr->data.resident.value_offset));
379	vol->vol_flags = vi->flags = flags;
380	flush_dcache_mft_record_page(ctx->ntfs_ino);
381	mark_mft_record_dirty(ctx->ntfs_ino);
382	ntfs_attr_put_search_ctx(ctx);
383	unmap_mft_record(ni);
384done:
385	ntfs_debug("Done.");
386	return 0;
387put_unm_err_out:
388	if (ctx)
389		ntfs_attr_put_search_ctx(ctx);
390	unmap_mft_record(ni);
391err_out:
392	ntfs_error(vol->sb, "Failed with error code %i.", -err);
393	return err;
394}
395
396/**
397 * ntfs_set_volume_flags - set bits in the volume information flags
398 * @vol:	ntfs volume on which to modify the flags
399 * @flags:	flags to set on the volume
400 *
401 * Set the bits in @flags in the volume information flags on the volume @vol.
402 *
403 * Return 0 on success and -errno on error.
404 */
405static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
406{
407	flags &= VOLUME_FLAGS_MASK;
408	return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
409}
410
411/**
412 * ntfs_clear_volume_flags - clear bits in the volume information flags
413 * @vol:	ntfs volume on which to modify the flags
414 * @flags:	flags to clear on the volume
415 *
416 * Clear the bits in @flags in the volume information flags on the volume @vol.
417 *
418 * Return 0 on success and -errno on error.
419 */
420static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
421{
422	flags &= VOLUME_FLAGS_MASK;
423	flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
424	return ntfs_write_volume_flags(vol, flags);
425}
426
427#endif /* NTFS_RW */
428
429/**
430 * ntfs_remount - change the mount options of a mounted ntfs filesystem
431 * @sb:		superblock of mounted ntfs filesystem
432 * @flags:	remount flags
433 * @opt:	remount options string
434 *
435 * Change the mount options of an already mounted ntfs filesystem.
436 *
437 * NOTE:  The VFS sets the @sb->s_flags remount flags to @flags after
438 * ntfs_remount() returns successfully (i.e. returns 0).  Otherwise,
439 * @sb->s_flags are not changed.
440 */
441static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
442{
443	ntfs_volume *vol = NTFS_SB(sb);
444
445	ntfs_debug("Entering with remount options string: %s", opt);
446#ifndef NTFS_RW
447	/* For read-only compiled driver, enforce read-only flag. */
448	*flags |= MS_RDONLY;
449#else /* NTFS_RW */
450	/*
451	 * For the read-write compiled driver, if we are remounting read-write,
452	 * make sure there are no volume errors and that no unsupported volume
453	 * flags are set.  Also, empty the logfile journal as it would become
454	 * stale as soon as something is written to the volume and mark the
455	 * volume dirty so that chkdsk is run if the volume is not umounted
456	 * cleanly.  Finally, mark the quotas out of date so Windows rescans
457	 * the volume on boot and updates them.
458	 *
459	 * When remounting read-only, mark the volume clean if no volume errors
460	 * have occured.
461	 */
462	if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
463		static const char *es = ".  Cannot remount read-write.";
464
465		/* Remounting read-write. */
466		if (NVolErrors(vol)) {
467			ntfs_error(sb, "Volume has errors and is read-only%s",
468					es);
469			return -EROFS;
470		}
471		if (vol->vol_flags & VOLUME_IS_DIRTY) {
472			ntfs_error(sb, "Volume is dirty and read-only%s", es);
473			return -EROFS;
474		}
475		if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
476			ntfs_error(sb, "Volume has been modified by chkdsk "
477					"and is read-only%s", es);
478			return -EROFS;
479		}
480		if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
481			ntfs_error(sb, "Volume has unsupported flags set "
482					"(0x%x) and is read-only%s",
483					(unsigned)le16_to_cpu(vol->vol_flags),
484					es);
485			return -EROFS;
486		}
487		if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
488			ntfs_error(sb, "Failed to set dirty bit in volume "
489					"information flags%s", es);
490			return -EROFS;
491		}
492		if (!ntfs_empty_logfile(vol->logfile_ino)) {
493			ntfs_error(sb, "Failed to empty journal $LogFile%s",
494					es);
495			NVolSetErrors(vol);
496			return -EROFS;
497		}
498		if (!ntfs_mark_quotas_out_of_date(vol)) {
499			ntfs_error(sb, "Failed to mark quotas out of date%s",
500					es);
501			NVolSetErrors(vol);
502			return -EROFS;
503		}
504		if (!ntfs_stamp_usnjrnl(vol)) {
505			ntfs_error(sb, "Failed to stamp transation log "
506					"($UsnJrnl)%s", es);
507			NVolSetErrors(vol);
508			return -EROFS;
509		}
510	} else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
511		/* Remounting read-only. */
512		if (!NVolErrors(vol)) {
513			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
514				ntfs_warning(sb, "Failed to clear dirty bit "
515						"in volume information "
516						"flags.  Run chkdsk.");
517		}
518	}
519#endif /* NTFS_RW */
520
521	// TODO: Deal with *flags.
522
523	if (!parse_options(vol, opt))
524		return -EINVAL;
525	ntfs_debug("Done.");
526	return 0;
527}
528
529/**
530 * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
531 * @sb:		Super block of the device to which @b belongs.
532 * @b:		Boot sector of device @sb to check.
533 * @silent:	If 'true', all output will be silenced.
534 *
535 * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
536 * sector. Returns 'true' if it is valid and 'false' if not.
537 *
538 * @sb is only needed for warning/error output, i.e. it can be NULL when silent
539 * is 'true'.
540 */
541static bool is_boot_sector_ntfs(const struct super_block *sb,
542		const NTFS_BOOT_SECTOR *b, const bool silent)
543{
544	/*
545	 * Check that checksum == sum of u32 values from b to the checksum
546	 * field.  If checksum is zero, no checking is done.  We will work when
547	 * the checksum test fails, since some utilities update the boot sector
548	 * ignoring the checksum which leaves the checksum out-of-date.  We
549	 * report a warning if this is the case.
550	 */
551	if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
552		le32 *u;
553		u32 i;
554
555		for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
556			i += le32_to_cpup(u);
557		if (le32_to_cpu(b->checksum) != i)
558			ntfs_warning(sb, "Invalid boot sector checksum.");
559	}
560	/* Check OEMidentifier is "NTFS    " */
561	if (b->oem_id != magicNTFS)
562		goto not_ntfs;
563	/* Check bytes per sector value is between 256 and 4096. */
564	if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
565			le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
566		goto not_ntfs;
567	/* Check sectors per cluster value is valid. */
568	switch (b->bpb.sectors_per_cluster) {
569	case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
570		break;
571	default:
572		goto not_ntfs;
573	}
574	/* Check the cluster size is not above the maximum (64kiB). */
575	if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
576			b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
577		goto not_ntfs;
578	/* Check reserved/unused fields are really zero. */
579	if (le16_to_cpu(b->bpb.reserved_sectors) ||
580			le16_to_cpu(b->bpb.root_entries) ||
581			le16_to_cpu(b->bpb.sectors) ||
582			le16_to_cpu(b->bpb.sectors_per_fat) ||
583			le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
584		goto not_ntfs;
585	/* Check clusters per file mft record value is valid. */
586	if ((u8)b->clusters_per_mft_record < 0xe1 ||
587			(u8)b->clusters_per_mft_record > 0xf7)
588		switch (b->clusters_per_mft_record) {
589		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
590			break;
591		default:
592			goto not_ntfs;
593		}
594	/* Check clusters per index block value is valid. */
595	if ((u8)b->clusters_per_index_record < 0xe1 ||
596			(u8)b->clusters_per_index_record > 0xf7)
597		switch (b->clusters_per_index_record) {
598		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
599			break;
600		default:
601			goto not_ntfs;
602		}
603	/*
604	 * Check for valid end of sector marker. We will work without it, but
605	 * many BIOSes will refuse to boot from a bootsector if the magic is
606	 * incorrect, so we emit a warning.
607	 */
608	if (!silent && b->end_of_sector_marker != const_cpu_to_le16(0xaa55))
609		ntfs_warning(sb, "Invalid end of sector marker.");
610	return true;
611not_ntfs:
612	return false;
613}
614
615/**
616 * read_ntfs_boot_sector - read the NTFS boot sector of a device
617 * @sb:		super block of device to read the boot sector from
618 * @silent:	if true, suppress all output
619 *
620 * Reads the boot sector from the device and validates it. If that fails, tries
621 * to read the backup boot sector, first from the end of the device a-la NT4 and
622 * later and then from the middle of the device a-la NT3.51 and before.
623 *
624 * If a valid boot sector is found but it is not the primary boot sector, we
625 * repair the primary boot sector silently (unless the device is read-only or
626 * the primary boot sector is not accessible).
627 *
628 * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
629 * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
630 * to their respective values.
631 *
632 * Return the unlocked buffer head containing the boot sector or NULL on error.
633 */
634static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
635		const int silent)
636{
637	const char *read_err_str = "Unable to read %s boot sector.";
638	struct buffer_head *bh_primary, *bh_backup;
639	sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
640
641	/* Try to read primary boot sector. */
642	if ((bh_primary = sb_bread(sb, 0))) {
643		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
644				bh_primary->b_data, silent))
645			return bh_primary;
646		if (!silent)
647			ntfs_error(sb, "Primary boot sector is invalid.");
648	} else if (!silent)
649		ntfs_error(sb, read_err_str, "primary");
650	if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
651		if (bh_primary)
652			brelse(bh_primary);
653		if (!silent)
654			ntfs_error(sb, "Mount option errors=recover not used. "
655					"Aborting without trying to recover.");
656		return NULL;
657	}
658	/* Try to read NT4+ backup boot sector. */
659	if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
660		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
661				bh_backup->b_data, silent))
662			goto hotfix_primary_boot_sector;
663		brelse(bh_backup);
664	} else if (!silent)
665		ntfs_error(sb, read_err_str, "backup");
666	/* Try to read NT3.51- backup boot sector. */
667	if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
668		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
669				bh_backup->b_data, silent))
670			goto hotfix_primary_boot_sector;
671		if (!silent)
672			ntfs_error(sb, "Could not find a valid backup boot "
673					"sector.");
674		brelse(bh_backup);
675	} else if (!silent)
676		ntfs_error(sb, read_err_str, "backup");
677	/* We failed. Cleanup and return. */
678	if (bh_primary)
679		brelse(bh_primary);
680	return NULL;
681hotfix_primary_boot_sector:
682	if (bh_primary) {
683		if (!(sb->s_flags & MS_RDONLY)) {
684			ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
685					"boot sector from backup copy.");
686			memcpy(bh_primary->b_data, bh_backup->b_data,
687					NTFS_BLOCK_SIZE);
688			mark_buffer_dirty(bh_primary);
689			sync_dirty_buffer(bh_primary);
690			if (buffer_uptodate(bh_primary)) {
691				brelse(bh_backup);
692				return bh_primary;
693			}
694			ntfs_error(sb, "Hot-fix: Device write error while "
695					"recovering primary boot sector.");
696		} else {
697			ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
698					"sector failed: Read-only mount.");
699		}
700		brelse(bh_primary);
701	}
702	ntfs_warning(sb, "Using backup boot sector.");
703	return bh_backup;
704}
705
706/**
707 * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
708 * @vol:	volume structure to initialise with data from boot sector
709 * @b:		boot sector to parse
710 *
711 * Parse the ntfs boot sector @b and store all imporant information therein in
712 * the ntfs super block @vol.  Return 'true' on success and 'false' on error.
713 */
714static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
715{
716	unsigned int sectors_per_cluster_bits, nr_hidden_sects;
717	int clusters_per_mft_record, clusters_per_index_record;
718	s64 ll;
719
720	vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
721	vol->sector_size_bits = ffs(vol->sector_size) - 1;
722	ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
723			vol->sector_size);
724	ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
725			vol->sector_size_bits);
726	if (vol->sector_size < vol->sb->s_blocksize) {
727		ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
728				"device block size (%lu).  This is not "
729				"supported.  Sorry.", vol->sector_size,
730				vol->sb->s_blocksize);
731		return false;
732	}
733	ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
734	sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
735	ntfs_debug("sectors_per_cluster_bits = 0x%x",
736			sectors_per_cluster_bits);
737	nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
738	ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
739	vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
740	vol->cluster_size_mask = vol->cluster_size - 1;
741	vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
742	ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
743			vol->cluster_size);
744	ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
745	ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
746	if (vol->cluster_size < vol->sector_size) {
747		ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
748				"sector size (%i).  This is not supported.  "
749				"Sorry.", vol->cluster_size, vol->sector_size);
750		return false;
751	}
752	clusters_per_mft_record = b->clusters_per_mft_record;
753	ntfs_debug("clusters_per_mft_record = %i (0x%x)",
754			clusters_per_mft_record, clusters_per_mft_record);
755	if (clusters_per_mft_record > 0)
756		vol->mft_record_size = vol->cluster_size <<
757				(ffs(clusters_per_mft_record) - 1);
758	else
759		/*
760		 * When mft_record_size < cluster_size, clusters_per_mft_record
761		 * = -log2(mft_record_size) bytes. mft_record_size normaly is
762		 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
763		 */
764		vol->mft_record_size = 1 << -clusters_per_mft_record;
765	vol->mft_record_size_mask = vol->mft_record_size - 1;
766	vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
767	ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
768			vol->mft_record_size);
769	ntfs_debug("vol->mft_record_size_mask = 0x%x",
770			vol->mft_record_size_mask);
771	ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
772			vol->mft_record_size_bits, vol->mft_record_size_bits);
773	/*
774	 * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
775	 * we store $MFT/$DATA, the table of mft records in the page cache.
776	 */
777	if (vol->mft_record_size > PAGE_CACHE_SIZE) {
778		ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
779				"PAGE_CACHE_SIZE on your system (%lu).  "
780				"This is not supported.  Sorry.",
781				vol->mft_record_size, PAGE_CACHE_SIZE);
782		return false;
783	}
784	/* We cannot support mft record sizes below the sector size. */
785	if (vol->mft_record_size < vol->sector_size) {
786		ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
787				"sector size (%i).  This is not supported.  "
788				"Sorry.", vol->mft_record_size,
789				vol->sector_size);
790		return false;
791	}
792	clusters_per_index_record = b->clusters_per_index_record;
793	ntfs_debug("clusters_per_index_record = %i (0x%x)",
794			clusters_per_index_record, clusters_per_index_record);
795	if (clusters_per_index_record > 0)
796		vol->index_record_size = vol->cluster_size <<
797				(ffs(clusters_per_index_record) - 1);
798	else
799		/*
800		 * When index_record_size < cluster_size,
801		 * clusters_per_index_record = -log2(index_record_size) bytes.
802		 * index_record_size normaly equals 4096 bytes, which is
803		 * encoded as 0xF4 (-12 in decimal).
804		 */
805		vol->index_record_size = 1 << -clusters_per_index_record;
806	vol->index_record_size_mask = vol->index_record_size - 1;
807	vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
808	ntfs_debug("vol->index_record_size = %i (0x%x)",
809			vol->index_record_size, vol->index_record_size);
810	ntfs_debug("vol->index_record_size_mask = 0x%x",
811			vol->index_record_size_mask);
812	ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
813			vol->index_record_size_bits,
814			vol->index_record_size_bits);
815	/* We cannot support index record sizes below the sector size. */
816	if (vol->index_record_size < vol->sector_size) {
817		ntfs_error(vol->sb, "Index record size (%i) is smaller than "
818				"the sector size (%i).  This is not "
819				"supported.  Sorry.", vol->index_record_size,
820				vol->sector_size);
821		return false;
822	}
823	/*
824	 * Get the size of the volume in clusters and check for 64-bit-ness.
825	 * Windows currently only uses 32 bits to save the clusters so we do
826	 * the same as it is much faster on 32-bit CPUs.
827	 */
828	ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
829	if ((u64)ll >= 1ULL << 32) {
830		ntfs_error(vol->sb, "Cannot handle 64-bit clusters.  Sorry.");
831		return false;
832	}
833	vol->nr_clusters = ll;
834	ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
835	/*
836	 * On an architecture where unsigned long is 32-bits, we restrict the
837	 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
838	 * will hopefully optimize the whole check away.
839	 */
840	if (sizeof(unsigned long) < 8) {
841		if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
842			ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
843					"large for this architecture.  "
844					"Maximum supported is 2TiB.  Sorry.",
845					(unsigned long long)ll >> (40 -
846					vol->cluster_size_bits));
847			return false;
848		}
849	}
850	ll = sle64_to_cpu(b->mft_lcn);
851	if (ll >= vol->nr_clusters) {
852		ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
853				"volume.  Weird.", (unsigned long long)ll,
854				(unsigned long long)ll);
855		return false;
856	}
857	vol->mft_lcn = ll;
858	ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
859	ll = sle64_to_cpu(b->mftmirr_lcn);
860	if (ll >= vol->nr_clusters) {
861		ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
862				"of volume.  Weird.", (unsigned long long)ll,
863				(unsigned long long)ll);
864		return false;
865	}
866	vol->mftmirr_lcn = ll;
867	ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
868#ifdef NTFS_RW
869	/*
870	 * Work out the size of the mft mirror in number of mft records. If the
871	 * cluster size is less than or equal to the size taken by four mft
872	 * records, the mft mirror stores the first four mft records. If the
873	 * cluster size is bigger than the size taken by four mft records, the
874	 * mft mirror contains as many mft records as will fit into one
875	 * cluster.
876	 */
877	if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
878		vol->mftmirr_size = 4;
879	else
880		vol->mftmirr_size = vol->cluster_size >>
881				vol->mft_record_size_bits;
882	ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
883#endif /* NTFS_RW */
884	vol->serial_no = le64_to_cpu(b->volume_serial_number);
885	ntfs_debug("vol->serial_no = 0x%llx",
886			(unsigned long long)vol->serial_no);
887	return true;
888}
889
890/**
891 * ntfs_setup_allocators - initialize the cluster and mft allocators
892 * @vol:	volume structure for which to setup the allocators
893 *
894 * Setup the cluster (lcn) and mft allocators to the starting values.
895 */
896static void ntfs_setup_allocators(ntfs_volume *vol)
897{
898#ifdef NTFS_RW
899	LCN mft_zone_size, mft_lcn;
900#endif /* NTFS_RW */
901
902	ntfs_debug("vol->mft_zone_multiplier = 0x%x",
903			vol->mft_zone_multiplier);
904#ifdef NTFS_RW
905	/* Determine the size of the MFT zone. */
906	mft_zone_size = vol->nr_clusters;
907	switch (vol->mft_zone_multiplier) {  /* % of volume size in clusters */
908	case 4:
909		mft_zone_size >>= 1;			/* 50%   */
910		break;
911	case 3:
912		mft_zone_size = (mft_zone_size +
913				(mft_zone_size >> 1)) >> 2;	/* 37.5% */
914		break;
915	case 2:
916		mft_zone_size >>= 2;			/* 25%   */
917		break;
918	/* case 1: */
919	default:
920		mft_zone_size >>= 3;			/* 12.5% */
921		break;
922	}
923	/* Setup the mft zone. */
924	vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
925	ntfs_debug("vol->mft_zone_pos = 0x%llx",
926			(unsigned long long)vol->mft_zone_pos);
927	/*
928	 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
929	 * source) and if the actual mft_lcn is in the expected place or even
930	 * further to the front of the volume, extend the mft_zone to cover the
931	 * beginning of the volume as well.  This is in order to protect the
932	 * area reserved for the mft bitmap as well within the mft_zone itself.
933	 * On non-standard volumes we do not protect it as the overhead would
934	 * be higher than the speed increase we would get by doing it.
935	 */
936	mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
937	if (mft_lcn * vol->cluster_size < 16 * 1024)
938		mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
939				vol->cluster_size;
940	if (vol->mft_zone_start <= mft_lcn)
941		vol->mft_zone_start = 0;
942	ntfs_debug("vol->mft_zone_start = 0x%llx",
943			(unsigned long long)vol->mft_zone_start);
944	/*
945	 * Need to cap the mft zone on non-standard volumes so that it does
946	 * not point outside the boundaries of the volume.  We do this by
947	 * halving the zone size until we are inside the volume.
948	 */
949	vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
950	while (vol->mft_zone_end >= vol->nr_clusters) {
951		mft_zone_size >>= 1;
952		vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
953	}
954	ntfs_debug("vol->mft_zone_end = 0x%llx",
955			(unsigned long long)vol->mft_zone_end);
956	/*
957	 * Set the current position within each data zone to the start of the
958	 * respective zone.
959	 */
960	vol->data1_zone_pos = vol->mft_zone_end;
961	ntfs_debug("vol->data1_zone_pos = 0x%llx",
962			(unsigned long long)vol->data1_zone_pos);
963	vol->data2_zone_pos = 0;
964	ntfs_debug("vol->data2_zone_pos = 0x%llx",
965			(unsigned long long)vol->data2_zone_pos);
966
967	/* Set the mft data allocation position to mft record 24. */
968	vol->mft_data_pos = 24;
969	ntfs_debug("vol->mft_data_pos = 0x%llx",
970			(unsigned long long)vol->mft_data_pos);
971#endif /* NTFS_RW */
972}
973
974#ifdef NTFS_RW
975
976/**
977 * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
978 * @vol:	ntfs super block describing device whose mft mirror to load
979 *
980 * Return 'true' on success or 'false' on error.
981 */
982static bool load_and_init_mft_mirror(ntfs_volume *vol)
983{
984	struct inode *tmp_ino;
985	ntfs_inode *tmp_ni;
986
987	ntfs_debug("Entering.");
988	/* Get mft mirror inode. */
989	tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
990	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
991		if (!IS_ERR(tmp_ino))
992			iput(tmp_ino);
993		/* Caller will display error message. */
994		return false;
995	}
996	/*
997	 * Re-initialize some specifics about $MFTMirr's inode as
998	 * ntfs_read_inode() will have set up the default ones.
999	 */
1000	/* Set uid and gid to root. */
1001	tmp_ino->i_uid = tmp_ino->i_gid = 0;
1002	/* Regular file.  No access for anyone. */
1003	tmp_ino->i_mode = S_IFREG;
1004	/* No VFS initiated operations allowed for $MFTMirr. */
1005	tmp_ino->i_op = &ntfs_empty_inode_ops;
1006	tmp_ino->i_fop = &ntfs_empty_file_ops;
1007	/* Put in our special address space operations. */
1008	tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1009	tmp_ni = NTFS_I(tmp_ino);
1010	/* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1011	NInoSetMstProtected(tmp_ni);
1012	NInoSetSparseDisabled(tmp_ni);
1013	/*
1014	 * Set up our little cheat allowing us to reuse the async read io
1015	 * completion handler for directories.
1016	 */
1017	tmp_ni->itype.index.block_size = vol->mft_record_size;
1018	tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1019	vol->mftmirr_ino = tmp_ino;
1020	ntfs_debug("Done.");
1021	return true;
1022}
1023
1024/**
1025 * check_mft_mirror - compare contents of the mft mirror with the mft
1026 * @vol:	ntfs super block describing device whose mft mirror to check
1027 *
1028 * Return 'true' on success or 'false' on error.
1029 *
1030 * Note, this function also results in the mft mirror runlist being completely
1031 * mapped into memory.  The mft mirror write code requires this and will BUG()
1032 * should it find an unmapped runlist element.
1033 */
1034static bool check_mft_mirror(ntfs_volume *vol)
1035{
1036	struct super_block *sb = vol->sb;
1037	ntfs_inode *mirr_ni;
1038	struct page *mft_page, *mirr_page;
1039	u8 *kmft, *kmirr;
1040	runlist_element *rl, rl2[2];
1041	pgoff_t index;
1042	int mrecs_per_page, i;
1043
1044	ntfs_debug("Entering.");
1045	/* Compare contents of $MFT and $MFTMirr. */
1046	mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1047	BUG_ON(!mrecs_per_page);
1048	BUG_ON(!vol->mftmirr_size);
1049	mft_page = mirr_page = NULL;
1050	kmft = kmirr = NULL;
1051	index = i = 0;
1052	do {
1053		u32 bytes;
1054
1055		/* Switch pages if necessary. */
1056		if (!(i % mrecs_per_page)) {
1057			if (index) {
1058				ntfs_unmap_page(mft_page);
1059				ntfs_unmap_page(mirr_page);
1060			}
1061			/* Get the $MFT page. */
1062			mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1063					index);
1064			if (IS_ERR(mft_page)) {
1065				ntfs_error(sb, "Failed to read $MFT.");
1066				return false;
1067			}
1068			kmft = page_address(mft_page);
1069			/* Get the $MFTMirr page. */
1070			mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1071					index);
1072			if (IS_ERR(mirr_page)) {
1073				ntfs_error(sb, "Failed to read $MFTMirr.");
1074				goto mft_unmap_out;
1075			}
1076			kmirr = page_address(mirr_page);
1077			++index;
1078		}
1079		/* Do not check the record if it is not in use. */
1080		if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1081			/* Make sure the record is ok. */
1082			if (ntfs_is_baad_recordp((le32*)kmft)) {
1083				ntfs_error(sb, "Incomplete multi sector "
1084						"transfer detected in mft "
1085						"record %i.", i);
1086mm_unmap_out:
1087				ntfs_unmap_page(mirr_page);
1088mft_unmap_out:
1089				ntfs_unmap_page(mft_page);
1090				return false;
1091			}
1092		}
1093		/* Do not check the mirror record if it is not in use. */
1094		if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1095			if (ntfs_is_baad_recordp((le32*)kmirr)) {
1096				ntfs_error(sb, "Incomplete multi sector "
1097						"transfer detected in mft "
1098						"mirror record %i.", i);
1099				goto mm_unmap_out;
1100			}
1101		}
1102		/* Get the amount of data in the current record. */
1103		bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1104		if (bytes < sizeof(MFT_RECORD_OLD) ||
1105				bytes > vol->mft_record_size ||
1106				ntfs_is_baad_recordp((le32*)kmft)) {
1107			bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1108			if (bytes < sizeof(MFT_RECORD_OLD) ||
1109					bytes > vol->mft_record_size ||
1110					ntfs_is_baad_recordp((le32*)kmirr))
1111				bytes = vol->mft_record_size;
1112		}
1113		/* Compare the two records. */
1114		if (memcmp(kmft, kmirr, bytes)) {
1115			ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1116					"match.  Run ntfsfix or chkdsk.", i);
1117			goto mm_unmap_out;
1118		}
1119		kmft += vol->mft_record_size;
1120		kmirr += vol->mft_record_size;
1121	} while (++i < vol->mftmirr_size);
1122	/* Release the last pages. */
1123	ntfs_unmap_page(mft_page);
1124	ntfs_unmap_page(mirr_page);
1125
1126	/* Construct the mft mirror runlist by hand. */
1127	rl2[0].vcn = 0;
1128	rl2[0].lcn = vol->mftmirr_lcn;
1129	rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1130			vol->cluster_size - 1) / vol->cluster_size;
1131	rl2[1].vcn = rl2[0].length;
1132	rl2[1].lcn = LCN_ENOENT;
1133	rl2[1].length = 0;
1134	/*
1135	 * Because we have just read all of the mft mirror, we know we have
1136	 * mapped the full runlist for it.
1137	 */
1138	mirr_ni = NTFS_I(vol->mftmirr_ino);
1139	down_read(&mirr_ni->runlist.lock);
1140	rl = mirr_ni->runlist.rl;
1141	/* Compare the two runlists.  They must be identical. */
1142	i = 0;
1143	do {
1144		if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1145				rl2[i].length != rl[i].length) {
1146			ntfs_error(sb, "$MFTMirr location mismatch.  "
1147					"Run chkdsk.");
1148			up_read(&mirr_ni->runlist.lock);
1149			return false;
1150		}
1151	} while (rl2[i++].length);
1152	up_read(&mirr_ni->runlist.lock);
1153	ntfs_debug("Done.");
1154	return true;
1155}
1156
1157/**
1158 * load_and_check_logfile - load and check the logfile inode for a volume
1159 * @vol:	ntfs super block describing device whose logfile to load
1160 *
1161 * Return 'true' on success or 'false' on error.
1162 */
1163static bool load_and_check_logfile(ntfs_volume *vol,
1164		RESTART_PAGE_HEADER **rp)
1165{
1166	struct inode *tmp_ino;
1167
1168	ntfs_debug("Entering.");
1169	tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1170	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1171		if (!IS_ERR(tmp_ino))
1172			iput(tmp_ino);
1173		/* Caller will display error message. */
1174		return false;
1175	}
1176	if (!ntfs_check_logfile(tmp_ino, rp)) {
1177		iput(tmp_ino);
1178		/* ntfs_check_logfile() will have displayed error output. */
1179		return false;
1180	}
1181	NInoSetSparseDisabled(NTFS_I(tmp_ino));
1182	vol->logfile_ino = tmp_ino;
1183	ntfs_debug("Done.");
1184	return true;
1185}
1186
1187#define NTFS_HIBERFIL_HEADER_SIZE	4096
1188
1189/**
1190 * check_windows_hibernation_status - check if Windows is suspended on a volume
1191 * @vol:	ntfs super block of device to check
1192 *
1193 * Check if Windows is hibernated on the ntfs volume @vol.  This is done by
1194 * looking for the file hiberfil.sys in the root directory of the volume.  If
1195 * the file is not present Windows is definitely not suspended.
1196 *
1197 * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1198 * definitely suspended (this volume is not the system volume).  Caveat:  on a
1199 * system with many volumes it is possible that the < 4kiB check is bogus but
1200 * for now this should do fine.
1201 *
1202 * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1203 * hiberfil header (which is the first 4kiB).  If this begins with "hibr",
1204 * Windows is definitely suspended.  If it is completely full of zeroes,
1205 * Windows is definitely not hibernated.  Any other case is treated as if
1206 * Windows is suspended.  This caters for the above mentioned caveat of a
1207 * system with many volumes where no "hibr" magic would be present and there is
1208 * no zero header.
1209 *
1210 * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1211 * hibernated on the volume, and -errno on error.
1212 */
1213static int check_windows_hibernation_status(ntfs_volume *vol)
1214{
1215	MFT_REF mref;
1216	struct inode *vi;
1217	ntfs_inode *ni;
1218	struct page *page;
1219	u32 *kaddr, *kend;
1220	ntfs_name *name = NULL;
1221	int ret = 1;
1222	static const ntfschar hiberfil[13] = { const_cpu_to_le16('h'),
1223			const_cpu_to_le16('i'), const_cpu_to_le16('b'),
1224			const_cpu_to_le16('e'), const_cpu_to_le16('r'),
1225			const_cpu_to_le16('f'), const_cpu_to_le16('i'),
1226			const_cpu_to_le16('l'), const_cpu_to_le16('.'),
1227			const_cpu_to_le16('s'), const_cpu_to_le16('y'),
1228			const_cpu_to_le16('s'), 0 };
1229
1230	ntfs_debug("Entering.");
1231	/*
1232	 * Find the inode number for the hibernation file by looking up the
1233	 * filename hiberfil.sys in the root directory.
1234	 */
1235	mutex_lock(&vol->root_ino->i_mutex);
1236	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1237			&name);
1238	mutex_unlock(&vol->root_ino->i_mutex);
1239	if (IS_ERR_MREF(mref)) {
1240		ret = MREF_ERR(mref);
1241		/* If the file does not exist, Windows is not hibernated. */
1242		if (ret == -ENOENT) {
1243			ntfs_debug("hiberfil.sys not present.  Windows is not "
1244					"hibernated on the volume.");
1245			return 0;
1246		}
1247		/* A real error occured. */
1248		ntfs_error(vol->sb, "Failed to find inode number for "
1249				"hiberfil.sys.");
1250		return ret;
1251	}
1252	/* We do not care for the type of match that was found. */
1253	kfree(name);
1254	/* Get the inode. */
1255	vi = ntfs_iget(vol->sb, MREF(mref));
1256	if (IS_ERR(vi) || is_bad_inode(vi)) {
1257		if (!IS_ERR(vi))
1258			iput(vi);
1259		ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1260		return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1261	}
1262	if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1263		ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx).  "
1264				"Windows is hibernated on the volume.  This "
1265				"is not the system volume.", i_size_read(vi));
1266		goto iput_out;
1267	}
1268	ni = NTFS_I(vi);
1269	page = ntfs_map_page(vi->i_mapping, 0);
1270	if (IS_ERR(page)) {
1271		ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1272		ret = PTR_ERR(page);
1273		goto iput_out;
1274	}
1275	kaddr = (u32*)page_address(page);
1276	if (*(le32*)kaddr == const_cpu_to_le32(0x72626968)/*'hibr'*/) {
1277		ntfs_debug("Magic \"hibr\" found in hiberfil.sys.  Windows is "
1278				"hibernated on the volume.  This is the "
1279				"system volume.");
1280		goto unm_iput_out;
1281	}
1282	kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1283	do {
1284		if (unlikely(*kaddr)) {
1285			ntfs_debug("hiberfil.sys is larger than 4kiB "
1286					"(0x%llx), does not contain the "
1287					"\"hibr\" magic, and does not have a "
1288					"zero header.  Windows is hibernated "
1289					"on the volume.  This is not the "
1290					"system volume.", i_size_read(vi));
1291			goto unm_iput_out;
1292		}
1293	} while (++kaddr < kend);
1294	ntfs_debug("hiberfil.sys contains a zero header.  Windows is not "
1295			"hibernated on the volume.  This is the system "
1296			"volume.");
1297	ret = 0;
1298unm_iput_out:
1299	ntfs_unmap_page(page);
1300iput_out:
1301	iput(vi);
1302	return ret;
1303}
1304
1305/**
1306 * load_and_init_quota - load and setup the quota file for a volume if present
1307 * @vol:	ntfs super block describing device whose quota file to load
1308 *
1309 * Return 'true' on success or 'false' on error.  If $Quota is not present, we
1310 * leave vol->quota_ino as NULL and return success.
1311 */
1312static bool load_and_init_quota(ntfs_volume *vol)
1313{
1314	MFT_REF mref;
1315	struct inode *tmp_ino;
1316	ntfs_name *name = NULL;
1317	static const ntfschar Quota[7] = { const_cpu_to_le16('$'),
1318			const_cpu_to_le16('Q'), const_cpu_to_le16('u'),
1319			const_cpu_to_le16('o'), const_cpu_to_le16('t'),
1320			const_cpu_to_le16('a'), 0 };
1321	static ntfschar Q[3] = { const_cpu_to_le16('$'),
1322			const_cpu_to_le16('Q'), 0 };
1323
1324	ntfs_debug("Entering.");
1325	/*
1326	 * Find the inode number for the quota file by looking up the filename
1327	 * $Quota in the extended system files directory $Extend.
1328	 */
1329	mutex_lock(&vol->extend_ino->i_mutex);
1330	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1331			&name);
1332	mutex_unlock(&vol->extend_ino->i_mutex);
1333	if (IS_ERR_MREF(mref)) {
1334		/*
1335		 * If the file does not exist, quotas are disabled and have
1336		 * never been enabled on this volume, just return success.
1337		 */
1338		if (MREF_ERR(mref) == -ENOENT) {
1339			ntfs_debug("$Quota not present.  Volume does not have "
1340					"quotas enabled.");
1341			/*
1342			 * No need to try to set quotas out of date if they are
1343			 * not enabled.
1344			 */
1345			NVolSetQuotaOutOfDate(vol);
1346			return true;
1347		}
1348		/* A real error occured. */
1349		ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1350		return false;
1351	}
1352	/* We do not care for the type of match that was found. */
1353	kfree(name);
1354	/* Get the inode. */
1355	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1356	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1357		if (!IS_ERR(tmp_ino))
1358			iput(tmp_ino);
1359		ntfs_error(vol->sb, "Failed to load $Quota.");
1360		return false;
1361	}
1362	vol->quota_ino = tmp_ino;
1363	/* Get the $Q index allocation attribute. */
1364	tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1365	if (IS_ERR(tmp_ino)) {
1366		ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1367		return false;
1368	}
1369	vol->quota_q_ino = tmp_ino;
1370	ntfs_debug("Done.");
1371	return true;
1372}
1373
1374/**
1375 * load_and_init_usnjrnl - load and setup the transaction log if present
1376 * @vol:	ntfs super block describing device whose usnjrnl file to load
1377 *
1378 * Return 'true' on success or 'false' on error.
1379 *
1380 * If $UsnJrnl is not present or in the process of being disabled, we set
1381 * NVolUsnJrnlStamped() and return success.
1382 *
1383 * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1384 * i.e. transaction logging has only just been enabled or the journal has been
1385 * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1386 * and return success.
1387 */
1388static bool load_and_init_usnjrnl(ntfs_volume *vol)
1389{
1390	MFT_REF mref;
1391	struct inode *tmp_ino;
1392	ntfs_inode *tmp_ni;
1393	struct page *page;
1394	ntfs_name *name = NULL;
1395	USN_HEADER *uh;
1396	static const ntfschar UsnJrnl[9] = { const_cpu_to_le16('$'),
1397			const_cpu_to_le16('U'), const_cpu_to_le16('s'),
1398			const_cpu_to_le16('n'), const_cpu_to_le16('J'),
1399			const_cpu_to_le16('r'), const_cpu_to_le16('n'),
1400			const_cpu_to_le16('l'), 0 };
1401	static ntfschar Max[5] = { const_cpu_to_le16('$'),
1402			const_cpu_to_le16('M'), const_cpu_to_le16('a'),
1403			const_cpu_to_le16('x'), 0 };
1404	static ntfschar J[3] = { const_cpu_to_le16('$'),
1405			const_cpu_to_le16('J'), 0 };
1406
1407	ntfs_debug("Entering.");
1408	/*
1409	 * Find the inode number for the transaction log file by looking up the
1410	 * filename $UsnJrnl in the extended system files directory $Extend.
1411	 */
1412	mutex_lock(&vol->extend_ino->i_mutex);
1413	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1414			&name);
1415	mutex_unlock(&vol->extend_ino->i_mutex);
1416	if (IS_ERR_MREF(mref)) {
1417		/*
1418		 * If the file does not exist, transaction logging is disabled,
1419		 * just return success.
1420		 */
1421		if (MREF_ERR(mref) == -ENOENT) {
1422			ntfs_debug("$UsnJrnl not present.  Volume does not "
1423					"have transaction logging enabled.");
1424not_enabled:
1425			/*
1426			 * No need to try to stamp the transaction log if
1427			 * transaction logging is not enabled.
1428			 */
1429			NVolSetUsnJrnlStamped(vol);
1430			return true;
1431		}
1432		/* A real error occured. */
1433		ntfs_error(vol->sb, "Failed to find inode number for "
1434				"$UsnJrnl.");
1435		return false;
1436	}
1437	/* We do not care for the type of match that was found. */
1438	kfree(name);
1439	/* Get the inode. */
1440	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1441	if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1442		if (!IS_ERR(tmp_ino))
1443			iput(tmp_ino);
1444		ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1445		return false;
1446	}
1447	vol->usnjrnl_ino = tmp_ino;
1448	/*
1449	 * If the transaction log is in the process of being deleted, we can
1450	 * ignore it.
1451	 */
1452	if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1453		ntfs_debug("$UsnJrnl in the process of being disabled.  "
1454				"Volume does not have transaction logging "
1455				"enabled.");
1456		goto not_enabled;
1457	}
1458	/* Get the $DATA/$Max attribute. */
1459	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1460	if (IS_ERR(tmp_ino)) {
1461		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1462				"attribute.");
1463		return false;
1464	}
1465	vol->usnjrnl_max_ino = tmp_ino;
1466	if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1467		ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1468				"attribute (size is 0x%llx but should be at "
1469				"least 0x%zx bytes).", i_size_read(tmp_ino),
1470				sizeof(USN_HEADER));
1471		return false;
1472	}
1473	/* Get the $DATA/$J attribute. */
1474	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1475	if (IS_ERR(tmp_ino)) {
1476		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1477				"attribute.");
1478		return false;
1479	}
1480	vol->usnjrnl_j_ino = tmp_ino;
1481	/* Verify $J is non-resident and sparse. */
1482	tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1483	if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1484		ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1485				"and/or not sparse.");
1486		return false;
1487	}
1488	/* Read the USN_HEADER from $DATA/$Max. */
1489	page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1490	if (IS_ERR(page)) {
1491		ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1492				"attribute.");
1493		return false;
1494	}
1495	uh = (USN_HEADER*)page_address(page);
1496	/* Sanity check the $Max. */
1497	if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1498			sle64_to_cpu(uh->maximum_size))) {
1499		ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1500				"maximum size (0x%llx).  $UsnJrnl is corrupt.",
1501				(long long)sle64_to_cpu(uh->allocation_delta),
1502				(long long)sle64_to_cpu(uh->maximum_size));
1503		ntfs_unmap_page(page);
1504		return false;
1505	}
1506	/*
1507	 * If the transaction log has been stamped and nothing has been written
1508	 * to it since, we do not need to stamp it.
1509	 */
1510	if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1511			i_size_read(vol->usnjrnl_j_ino))) {
1512		if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1513				i_size_read(vol->usnjrnl_j_ino))) {
1514			ntfs_unmap_page(page);
1515			ntfs_debug("$UsnJrnl is enabled but nothing has been "
1516					"logged since it was last stamped.  "
1517					"Treating this as if the volume does "
1518					"not have transaction logging "
1519					"enabled.");
1520			goto not_enabled;
1521		}
1522		ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1523				"which is out of bounds (0x%llx).  $UsnJrnl "
1524				"is corrupt.",
1525				(long long)sle64_to_cpu(uh->lowest_valid_usn),
1526				i_size_read(vol->usnjrnl_j_ino));
1527		ntfs_unmap_page(page);
1528		return false;
1529	}
1530	ntfs_unmap_page(page);
1531	ntfs_debug("Done.");
1532	return true;
1533}
1534
1535/**
1536 * load_and_init_attrdef - load the attribute definitions table for a volume
1537 * @vol:	ntfs super block describing device whose attrdef to load
1538 *
1539 * Return 'true' on success or 'false' on error.
1540 */
1541static bool load_and_init_attrdef(ntfs_volume *vol)
1542{
1543	loff_t i_size;
1544	struct super_block *sb = vol->sb;
1545	struct inode *ino;
1546	struct page *page;
1547	pgoff_t index, max_index;
1548	unsigned int size;
1549
1550	ntfs_debug("Entering.");
1551	/* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1552	ino = ntfs_iget(sb, FILE_AttrDef);
1553	if (IS_ERR(ino) || is_bad_inode(ino)) {
1554		if (!IS_ERR(ino))
1555			iput(ino);
1556		goto failed;
1557	}
1558	NInoSetSparseDisabled(NTFS_I(ino));
1559	/* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1560	i_size = i_size_read(ino);
1561	if (i_size <= 0 || i_size > 0x7fffffff)
1562		goto iput_failed;
1563	vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1564	if (!vol->attrdef)
1565		goto iput_failed;
1566	index = 0;
1567	max_index = i_size >> PAGE_CACHE_SHIFT;
1568	size = PAGE_CACHE_SIZE;
1569	while (index < max_index) {
1570		/* Read the attrdef table and copy it into the linear buffer. */
1571read_partial_attrdef_page:
1572		page = ntfs_map_page(ino->i_mapping, index);
1573		if (IS_ERR(page))
1574			goto free_iput_failed;
1575		memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1576				page_address(page), size);
1577		ntfs_unmap_page(page);
1578	};
1579	if (size == PAGE_CACHE_SIZE) {
1580		size = i_size & ~PAGE_CACHE_MASK;
1581		if (size)
1582			goto read_partial_attrdef_page;
1583	}
1584	vol->attrdef_size = i_size;
1585	ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1586	iput(ino);
1587	return true;
1588free_iput_failed:
1589	ntfs_free(vol->attrdef);
1590	vol->attrdef = NULL;
1591iput_failed:
1592	iput(ino);
1593failed:
1594	ntfs_error(sb, "Failed to initialize attribute definition table.");
1595	return false;
1596}
1597
1598#endif /* NTFS_RW */
1599
1600/**
1601 * load_and_init_upcase - load the upcase table for an ntfs volume
1602 * @vol:	ntfs super block describing device whose upcase to load
1603 *
1604 * Return 'true' on success or 'false' on error.
1605 */
1606static bool load_and_init_upcase(ntfs_volume *vol)
1607{
1608	loff_t i_size;
1609	struct super_block *sb = vol->sb;
1610	struct inode *ino;
1611	struct page *page;
1612	pgoff_t index, max_index;
1613	unsigned int size;
1614	int i, max;
1615
1616	ntfs_debug("Entering.");
1617	/* Read upcase table and setup vol->upcase and vol->upcase_len. */
1618	ino = ntfs_iget(sb, FILE_UpCase);
1619	if (IS_ERR(ino) || is_bad_inode(ino)) {
1620		if (!IS_ERR(ino))
1621			iput(ino);
1622		goto upcase_failed;
1623	}
1624	/*
1625	 * The upcase size must not be above 64k Unicode characters, must not
1626	 * be zero and must be a multiple of sizeof(ntfschar).
1627	 */
1628	i_size = i_size_read(ino);
1629	if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1630			i_size > 64ULL * 1024 * sizeof(ntfschar))
1631		goto iput_upcase_failed;
1632	vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1633	if (!vol->upcase)
1634		goto iput_upcase_failed;
1635	index = 0;
1636	max_index = i_size >> PAGE_CACHE_SHIFT;
1637	size = PAGE_CACHE_SIZE;
1638	while (index < max_index) {
1639		/* Read the upcase table and copy it into the linear buffer. */
1640read_partial_upcase_page:
1641		page = ntfs_map_page(ino->i_mapping, index);
1642		if (IS_ERR(page))
1643			goto iput_upcase_failed;
1644		memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1645				page_address(page), size);
1646		ntfs_unmap_page(page);
1647	};
1648	if (size == PAGE_CACHE_SIZE) {
1649		size = i_size & ~PAGE_CACHE_MASK;
1650		if (size)
1651			goto read_partial_upcase_page;
1652	}
1653	vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1654	ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1655			i_size, 64 * 1024 * sizeof(ntfschar));
1656	iput(ino);
1657	mutex_lock(&ntfs_lock);
1658	if (!default_upcase) {
1659		ntfs_debug("Using volume specified $UpCase since default is "
1660				"not present.");
1661		mutex_unlock(&ntfs_lock);
1662		return true;
1663	}
1664	max = default_upcase_len;
1665	if (max > vol->upcase_len)
1666		max = vol->upcase_len;
1667	for (i = 0; i < max; i++)
1668		if (vol->upcase[i] != default_upcase[i])
1669			break;
1670	if (i == max) {
1671		ntfs_free(vol->upcase);
1672		vol->upcase = default_upcase;
1673		vol->upcase_len = max;
1674		ntfs_nr_upcase_users++;
1675		mutex_unlock(&ntfs_lock);
1676		ntfs_debug("Volume specified $UpCase matches default. Using "
1677				"default.");
1678		return true;
1679	}
1680	mutex_unlock(&ntfs_lock);
1681	ntfs_debug("Using volume specified $UpCase since it does not match "
1682			"the default.");
1683	return true;
1684iput_upcase_failed:
1685	iput(ino);
1686	ntfs_free(vol->upcase);
1687	vol->upcase = NULL;
1688upcase_failed:
1689	mutex_lock(&ntfs_lock);
1690	if (default_upcase) {
1691		vol->upcase = default_upcase;
1692		vol->upcase_len = default_upcase_len;
1693		ntfs_nr_upcase_users++;
1694		mutex_unlock(&ntfs_lock);
1695		ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1696				"default.");
1697		return true;
1698	}
1699	mutex_unlock(&ntfs_lock);
1700	ntfs_error(sb, "Failed to initialize upcase table.");
1701	return false;
1702}
1703
1704/*
1705 * The lcn and mft bitmap inodes are NTFS-internal inodes with
1706 * their own special locking rules:
1707 */
1708static struct lock_class_key
1709	lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1710	mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1711
1712/**
1713 * load_system_files - open the system files using normal functions
1714 * @vol:	ntfs super block describing device whose system files to load
1715 *
1716 * Open the system files with normal access functions and complete setting up
1717 * the ntfs super block @vol.
1718 *
1719 * Return 'true' on success or 'false' on error.
1720 */
1721static bool load_system_files(ntfs_volume *vol)
1722{
1723	struct super_block *sb = vol->sb;
1724	MFT_RECORD *m;
1725	VOLUME_INFORMATION *vi;
1726	ntfs_attr_search_ctx *ctx;
1727#ifdef NTFS_RW
1728	RESTART_PAGE_HEADER *rp;
1729	int err;
1730#endif /* NTFS_RW */
1731
1732	ntfs_debug("Entering.");
1733#ifdef NTFS_RW
1734	/* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1735	if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1736		static const char *es1 = "Failed to load $MFTMirr";
1737		static const char *es2 = "$MFTMirr does not match $MFT";
1738		static const char *es3 = ".  Run ntfsfix and/or chkdsk.";
1739
1740		/* If a read-write mount, convert it to a read-only mount. */
1741		if (!(sb->s_flags & MS_RDONLY)) {
1742			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1743					ON_ERRORS_CONTINUE))) {
1744				ntfs_error(sb, "%s and neither on_errors="
1745						"continue nor on_errors="
1746						"remount-ro was specified%s",
1747						!vol->mftmirr_ino ? es1 : es2,
1748						es3);
1749				goto iput_mirr_err_out;
1750			}
1751			sb->s_flags |= MS_RDONLY;
1752			ntfs_error(sb, "%s.  Mounting read-only%s",
1753					!vol->mftmirr_ino ? es1 : es2, es3);
1754		} else
1755			ntfs_warning(sb, "%s.  Will not be able to remount "
1756					"read-write%s",
1757					!vol->mftmirr_ino ? es1 : es2, es3);
1758		/* This will prevent a read-write remount. */
1759		NVolSetErrors(vol);
1760	}
1761#endif /* NTFS_RW */
1762	/* Get mft bitmap attribute inode. */
1763	vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1764	if (IS_ERR(vol->mftbmp_ino)) {
1765		ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1766		goto iput_mirr_err_out;
1767	}
1768	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1769			   &mftbmp_runlist_lock_key);
1770	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1771			   &mftbmp_mrec_lock_key);
1772	/* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1773	if (!load_and_init_upcase(vol))
1774		goto iput_mftbmp_err_out;
1775#ifdef NTFS_RW
1776	/*
1777	 * Read attribute definitions table and setup @vol->attrdef and
1778	 * @vol->attrdef_size.
1779	 */
1780	if (!load_and_init_attrdef(vol))
1781		goto iput_upcase_err_out;
1782#endif /* NTFS_RW */
1783	/*
1784	 * Get the cluster allocation bitmap inode and verify the size, no
1785	 * need for any locking at this stage as we are already running
1786	 * exclusively as we are mount in progress task.
1787	 */
1788	vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1789	if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1790		if (!IS_ERR(vol->lcnbmp_ino))
1791			iput(vol->lcnbmp_ino);
1792		goto bitmap_failed;
1793	}
1794	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1795			   &lcnbmp_runlist_lock_key);
1796	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1797			   &lcnbmp_mrec_lock_key);
1798
1799	NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1800	if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1801		iput(vol->lcnbmp_ino);
1802bitmap_failed:
1803		ntfs_error(sb, "Failed to load $Bitmap.");
1804		goto iput_attrdef_err_out;
1805	}
1806	/*
1807	 * Get the volume inode and setup our cache of the volume flags and
1808	 * version.
1809	 */
1810	vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1811	if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1812		if (!IS_ERR(vol->vol_ino))
1813			iput(vol->vol_ino);
1814volume_failed:
1815		ntfs_error(sb, "Failed to load $Volume.");
1816		goto iput_lcnbmp_err_out;
1817	}
1818	m = map_mft_record(NTFS_I(vol->vol_ino));
1819	if (IS_ERR(m)) {
1820iput_volume_failed:
1821		iput(vol->vol_ino);
1822		goto volume_failed;
1823	}
1824	if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1825		ntfs_error(sb, "Failed to get attribute search context.");
1826		goto get_ctx_vol_failed;
1827	}
1828	if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1829			ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1830err_put_vol:
1831		ntfs_attr_put_search_ctx(ctx);
1832get_ctx_vol_failed:
1833		unmap_mft_record(NTFS_I(vol->vol_ino));
1834		goto iput_volume_failed;
1835	}
1836	vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1837			le16_to_cpu(ctx->attr->data.resident.value_offset));
1838	/* Some bounds checks. */
1839	if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1840			le32_to_cpu(ctx->attr->data.resident.value_length) >
1841			(u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1842		goto err_put_vol;
1843	/* Copy the volume flags and version to the ntfs_volume structure. */
1844	vol->vol_flags = vi->flags;
1845	vol->major_ver = vi->major_ver;
1846	vol->minor_ver = vi->minor_ver;
1847	ntfs_attr_put_search_ctx(ctx);
1848	unmap_mft_record(NTFS_I(vol->vol_ino));
1849	printk(KERN_INFO "NTFS volume version %i.%i.\n", vol->major_ver,
1850			vol->minor_ver);
1851	if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1852		ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1853				"volume version %i.%i (need at least version "
1854				"3.0).", vol->major_ver, vol->minor_ver);
1855		NVolClearSparseEnabled(vol);
1856	}
1857#ifdef NTFS_RW
1858	/* Make sure that no unsupported volume flags are set. */
1859	if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1860		static const char *es1a = "Volume is dirty";
1861		static const char *es1b = "Volume has been modified by chkdsk";
1862		static const char *es1c = "Volume has unsupported flags set";
1863		static const char *es2a = ".  Run chkdsk and mount in Windows.";
1864		static const char *es2b = ".  Mount in Windows.";
1865		const char *es1, *es2;
1866
1867		es2 = es2a;
1868		if (vol->vol_flags & VOLUME_IS_DIRTY)
1869			es1 = es1a;
1870		else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1871			es1 = es1b;
1872			es2 = es2b;
1873		} else {
1874			es1 = es1c;
1875			ntfs_warning(sb, "Unsupported volume flags 0x%x "
1876					"encountered.",
1877					(unsigned)le16_to_cpu(vol->vol_flags));
1878		}
1879		/* If a read-write mount, convert it to a read-only mount. */
1880		if (!(sb->s_flags & MS_RDONLY)) {
1881			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1882					ON_ERRORS_CONTINUE))) {
1883				ntfs_error(sb, "%s and neither on_errors="
1884						"continue nor on_errors="
1885						"remount-ro was specified%s",
1886						es1, es2);
1887				goto iput_vol_err_out;
1888			}
1889			sb->s_flags |= MS_RDONLY;
1890			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1891		} else
1892			ntfs_warning(sb, "%s.  Will not be able to remount "
1893					"read-write%s", es1, es2);
1894		/*
1895		 * Do not set NVolErrors() because ntfs_remount() re-checks the
1896		 * flags which we need to do in case any flags have changed.
1897		 */
1898	}
1899	/*
1900	 * Get the inode for the logfile, check it and determine if the volume
1901	 * was shutdown cleanly.
1902	 */
1903	rp = NULL;
1904	if (!load_and_check_logfile(vol, &rp) ||
1905			!ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1906		static const char *es1a = "Failed to load $LogFile";
1907		static const char *es1b = "$LogFile is not clean";
1908		static const char *es2 = ".  Mount in Windows.";
1909		const char *es1;
1910
1911		es1 = !vol->logfile_ino ? es1a : es1b;
1912		/* If a read-write mount, convert it to a read-only mount. */
1913		if (!(sb->s_flags & MS_RDONLY)) {
1914			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1915					ON_ERRORS_CONTINUE))) {
1916				ntfs_error(sb, "%s and neither on_errors="
1917						"continue nor on_errors="
1918						"remount-ro was specified%s",
1919						es1, es2);
1920				if (vol->logfile_ino) {
1921					BUG_ON(!rp);
1922					ntfs_free(rp);
1923				}
1924				goto iput_logfile_err_out;
1925			}
1926			sb->s_flags |= MS_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		/* This will prevent a read-write remount. */
1932		NVolSetErrors(vol);
1933	}
1934	ntfs_free(rp);
1935#endif /* NTFS_RW */
1936	/* Get the root directory inode so we can do path lookups. */
1937	vol->root_ino = ntfs_iget(sb, FILE_root);
1938	if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1939		if (!IS_ERR(vol->root_ino))
1940			iput(vol->root_ino);
1941		ntfs_error(sb, "Failed to load root directory.");
1942		goto iput_logfile_err_out;
1943	}
1944#ifdef NTFS_RW
1945	/*
1946	 * Check if Windows is suspended to disk on the target volume.  If it
1947	 * is hibernated, we must not write *anything* to the disk so set
1948	 * NVolErrors() without setting the dirty volume flag and mount
1949	 * read-only.  This will prevent read-write remounting and it will also
1950	 * prevent all writes.
1951	 */
1952	err = check_windows_hibernation_status(vol);
1953	if (unlikely(err)) {
1954		static const char *es1a = "Failed to determine if Windows is "
1955				"hibernated";
1956		static const char *es1b = "Windows is hibernated";
1957		static const char *es2 = ".  Run chkdsk.";
1958		const char *es1;
1959
1960		es1 = err < 0 ? es1a : es1b;
1961		/* If a read-write mount, convert it to a read-only mount. */
1962		if (!(sb->s_flags & MS_RDONLY)) {
1963			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1964					ON_ERRORS_CONTINUE))) {
1965				ntfs_error(sb, "%s and neither on_errors="
1966						"continue nor on_errors="
1967						"remount-ro was specified%s",
1968						es1, es2);
1969				goto iput_root_err_out;
1970			}
1971			sb->s_flags |= MS_RDONLY;
1972			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1973		} else
1974			ntfs_warning(sb, "%s.  Will not be able to remount "
1975					"read-write%s", es1, es2);
1976		/* This will prevent a read-write remount. */
1977		NVolSetErrors(vol);
1978	}
1979	/* If (still) a read-write mount, mark the volume dirty. */
1980	if (!(sb->s_flags & MS_RDONLY) &&
1981			ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
1982		static const char *es1 = "Failed to set dirty bit in volume "
1983				"information flags";
1984		static const char *es2 = ".  Run chkdsk.";
1985
1986		/* Convert to a read-only mount. */
1987		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1988				ON_ERRORS_CONTINUE))) {
1989			ntfs_error(sb, "%s and neither on_errors=continue nor "
1990					"on_errors=remount-ro was specified%s",
1991					es1, es2);
1992			goto iput_root_err_out;
1993		}
1994		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1995		sb->s_flags |= MS_RDONLY;
1996		/*
1997		 * Do not set NVolErrors() because ntfs_remount() might manage
1998		 * to set the dirty flag in which case all would be well.
1999		 */
2000	}
2001	/* If (still) a read-write mount, empty the logfile. */
2002	if (!(sb->s_flags & MS_RDONLY) &&
2003			!ntfs_empty_logfile(vol->logfile_ino)) {
2004		static const char *es1 = "Failed to empty $LogFile";
2005		static const char *es2 = ".  Mount in Windows.";
2006
2007		/* Convert to a read-only mount. */
2008		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2009				ON_ERRORS_CONTINUE))) {
2010			ntfs_error(sb, "%s and neither on_errors=continue nor "
2011					"on_errors=remount-ro was specified%s",
2012					es1, es2);
2013			goto iput_root_err_out;
2014		}
2015		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2016		sb->s_flags |= MS_RDONLY;
2017		NVolSetErrors(vol);
2018	}
2019#endif /* NTFS_RW */
2020	/* If on NTFS versions before 3.0, we are done. */
2021	if (unlikely(vol->major_ver < 3))
2022		return true;
2023	/* NTFS 3.0+ specific initialization. */
2024	/* Get the security descriptors inode. */
2025	vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2026	if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2027		if (!IS_ERR(vol->secure_ino))
2028			iput(vol->secure_ino);
2029		ntfs_error(sb, "Failed to load $Secure.");
2030		goto iput_root_err_out;
2031	}
2032	// TODO: Initialize security.
2033	/* Get the extended system files' directory inode. */
2034	vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2035	if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2036		if (!IS_ERR(vol->extend_ino))
2037			iput(vol->extend_ino);
2038		ntfs_error(sb, "Failed to load $Extend.");
2039		goto iput_sec_err_out;
2040	}
2041#ifdef NTFS_RW
2042	/* Find the quota file, load it if present, and set it up. */
2043	if (!load_and_init_quota(vol)) {
2044		static const char *es1 = "Failed to load $Quota";
2045		static const char *es2 = ".  Run chkdsk.";
2046
2047		/* If a read-write mount, convert it to a read-only mount. */
2048		if (!(sb->s_flags & MS_RDONLY)) {
2049			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2050					ON_ERRORS_CONTINUE))) {
2051				ntfs_error(sb, "%s and neither on_errors="
2052						"continue nor on_errors="
2053						"remount-ro was specified%s",
2054						es1, es2);
2055				goto iput_quota_err_out;
2056			}
2057			sb->s_flags |= MS_RDONLY;
2058			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2059		} else
2060			ntfs_warning(sb, "%s.  Will not be able to remount "
2061					"read-write%s", es1, es2);
2062		/* This will prevent a read-write remount. */
2063		NVolSetErrors(vol);
2064	}
2065	/* If (still) a read-write mount, mark the quotas out of date. */
2066	if (!(sb->s_flags & MS_RDONLY) &&
2067			!ntfs_mark_quotas_out_of_date(vol)) {
2068		static const char *es1 = "Failed to mark quotas out of date";
2069		static const char *es2 = ".  Run chkdsk.";
2070
2071		/* Convert to a read-only mount. */
2072		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2073				ON_ERRORS_CONTINUE))) {
2074			ntfs_error(sb, "%s and neither on_errors=continue nor "
2075					"on_errors=remount-ro was specified%s",
2076					es1, es2);
2077			goto iput_quota_err_out;
2078		}
2079		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2080		sb->s_flags |= MS_RDONLY;
2081		NVolSetErrors(vol);
2082	}
2083	/*
2084	 * Find the transaction log file ($UsnJrnl), load it if present, check
2085	 * it, and set it up.
2086	 */
2087	if (!load_and_init_usnjrnl(vol)) {
2088		static const char *es1 = "Failed to load $UsnJrnl";
2089		static const char *es2 = ".  Run chkdsk.";
2090
2091		/* If a read-write mount, convert it to a read-only mount. */
2092		if (!(sb->s_flags & MS_RDONLY)) {
2093			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2094					ON_ERRORS_CONTINUE))) {
2095				ntfs_error(sb, "%s and neither on_errors="
2096						"continue nor on_errors="
2097						"remount-ro was specified%s",
2098						es1, es2);
2099				goto iput_usnjrnl_err_out;
2100			}
2101			sb->s_flags |= MS_RDONLY;
2102			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2103		} else
2104			ntfs_warning(sb, "%s.  Will not be able to remount "
2105					"read-write%s", es1, es2);
2106		/* This will prevent a read-write remount. */
2107		NVolSetErrors(vol);
2108	}
2109	/* If (still) a read-write mount, stamp the transaction log. */
2110	if (!(sb->s_flags & MS_RDONLY) && !ntfs_stamp_usnjrnl(vol)) {
2111		static const char *es1 = "Failed to stamp transaction log "
2112				"($UsnJrnl)";
2113		static const char *es2 = ".  Run chkdsk.";
2114
2115		/* Convert to a read-only mount. */
2116		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2117				ON_ERRORS_CONTINUE))) {
2118			ntfs_error(sb, "%s and neither on_errors=continue nor "
2119					"on_errors=remount-ro was specified%s",
2120					es1, es2);
2121			goto iput_usnjrnl_err_out;
2122		}
2123		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2124		sb->s_flags |= MS_RDONLY;
2125		NVolSetErrors(vol);
2126	}
2127#endif /* NTFS_RW */
2128	return true;
2129#ifdef NTFS_RW
2130iput_usnjrnl_err_out:
2131	if (vol->usnjrnl_j_ino)
2132		iput(vol->usnjrnl_j_ino);
2133	if (vol->usnjrnl_max_ino)
2134		iput(vol->usnjrnl_max_ino);
2135	if (vol->usnjrnl_ino)
2136		iput(vol->usnjrnl_ino);
2137iput_quota_err_out:
2138	if (vol->quota_q_ino)
2139		iput(vol->quota_q_ino);
2140	if (vol->quota_ino)
2141		iput(vol->quota_ino);
2142	iput(vol->extend_ino);
2143#endif /* NTFS_RW */
2144iput_sec_err_out:
2145	iput(vol->secure_ino);
2146iput_root_err_out:
2147	iput(vol->root_ino);
2148iput_logfile_err_out:
2149#ifdef NTFS_RW
2150	if (vol->logfile_ino)
2151		iput(vol->logfile_ino);
2152iput_vol_err_out:
2153#endif /* NTFS_RW */
2154	iput(vol->vol_ino);
2155iput_lcnbmp_err_out:
2156	iput(vol->lcnbmp_ino);
2157iput_attrdef_err_out:
2158	vol->attrdef_size = 0;
2159	if (vol->attrdef) {
2160		ntfs_free(vol->attrdef);
2161		vol->attrdef = NULL;
2162	}
2163#ifdef NTFS_RW
2164iput_upcase_err_out:
2165#endif /* NTFS_RW */
2166	vol->upcase_len = 0;
2167	mutex_lock(&ntfs_lock);
2168	if (vol->upcase == default_upcase) {
2169		ntfs_nr_upcase_users--;
2170		vol->upcase = NULL;
2171	}
2172	mutex_unlock(&ntfs_lock);
2173	if (vol->upcase) {
2174		ntfs_free(vol->upcase);
2175		vol->upcase = NULL;
2176	}
2177iput_mftbmp_err_out:
2178	iput(vol->mftbmp_ino);
2179iput_mirr_err_out:
2180#ifdef NTFS_RW
2181	if (vol->mftmirr_ino)
2182		iput(vol->mftmirr_ino);
2183#endif /* NTFS_RW */
2184	return false;
2185}
2186
2187/**
2188 * ntfs_put_super - called by the vfs to unmount a volume
2189 * @sb:		vfs superblock of volume to unmount
2190 *
2191 * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2192 * the volume is being unmounted (umount system call has been invoked) and it
2193 * releases all inodes and memory belonging to the NTFS specific part of the
2194 * super block.
2195 */
2196static void ntfs_put_super(struct super_block *sb)
2197{
2198	ntfs_volume *vol = NTFS_SB(sb);
2199
2200	ntfs_debug("Entering.");
2201#ifdef NTFS_RW
2202	/*
2203	 * Commit all inodes while they are still open in case some of them
2204	 * cause others to be dirtied.
2205	 */
2206	ntfs_commit_inode(vol->vol_ino);
2207
2208	/* NTFS 3.0+ specific. */
2209	if (vol->major_ver >= 3) {
2210		if (vol->usnjrnl_j_ino)
2211			ntfs_commit_inode(vol->usnjrnl_j_ino);
2212		if (vol->usnjrnl_max_ino)
2213			ntfs_commit_inode(vol->usnjrnl_max_ino);
2214		if (vol->usnjrnl_ino)
2215			ntfs_commit_inode(vol->usnjrnl_ino);
2216		if (vol->quota_q_ino)
2217			ntfs_commit_inode(vol->quota_q_ino);
2218		if (vol->quota_ino)
2219			ntfs_commit_inode(vol->quota_ino);
2220		if (vol->extend_ino)
2221			ntfs_commit_inode(vol->extend_ino);
2222		if (vol->secure_ino)
2223			ntfs_commit_inode(vol->secure_ino);
2224	}
2225
2226	ntfs_commit_inode(vol->root_ino);
2227
2228	down_write(&vol->lcnbmp_lock);
2229	ntfs_commit_inode(vol->lcnbmp_ino);
2230	up_write(&vol->lcnbmp_lock);
2231
2232	down_write(&vol->mftbmp_lock);
2233	ntfs_commit_inode(vol->mftbmp_ino);
2234	up_write(&vol->mftbmp_lock);
2235
2236	if (vol->logfile_ino)
2237		ntfs_commit_inode(vol->logfile_ino);
2238
2239	if (vol->mftmirr_ino)
2240		ntfs_commit_inode(vol->mftmirr_ino);
2241	ntfs_commit_inode(vol->mft_ino);
2242
2243	/*
2244	 * If a read-write mount and no volume errors have occured, mark the
2245	 * volume clean.  Also, re-commit all affected inodes.
2246	 */
2247	if (!(sb->s_flags & MS_RDONLY)) {
2248		if (!NVolErrors(vol)) {
2249			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2250				ntfs_warning(sb, "Failed to clear dirty bit "
2251						"in volume information "
2252						"flags.  Run chkdsk.");
2253			ntfs_commit_inode(vol->vol_ino);
2254			ntfs_commit_inode(vol->root_ino);
2255			if (vol->mftmirr_ino)
2256				ntfs_commit_inode(vol->mftmirr_ino);
2257			ntfs_commit_inode(vol->mft_ino);
2258		} else {
2259			ntfs_warning(sb, "Volume has errors.  Leaving volume "
2260					"marked dirty.  Run chkdsk.");
2261		}
2262	}
2263#endif /* NTFS_RW */
2264
2265	iput(vol->vol_ino);
2266	vol->vol_ino = NULL;
2267
2268	/* NTFS 3.0+ specific clean up. */
2269	if (vol->major_ver >= 3) {
2270#ifdef NTFS_RW
2271		if (vol->usnjrnl_j_ino) {
2272			iput(vol->usnjrnl_j_ino);
2273			vol->usnjrnl_j_ino = NULL;
2274		}
2275		if (vol->usnjrnl_max_ino) {
2276			iput(vol->usnjrnl_max_ino);
2277			vol->usnjrnl_max_ino = NULL;
2278		}
2279		if (vol->usnjrnl_ino) {
2280			iput(vol->usnjrnl_ino);
2281			vol->usnjrnl_ino = NULL;
2282		}
2283		if (vol->quota_q_ino) {
2284			iput(vol->quota_q_ino);
2285			vol->quota_q_ino = NULL;
2286		}
2287		if (vol->quota_ino) {
2288			iput(vol->quota_ino);
2289			vol->quota_ino = NULL;
2290		}
2291#endif /* NTFS_RW */
2292		if (vol->extend_ino) {
2293			iput(vol->extend_ino);
2294			vol->extend_ino = NULL;
2295		}
2296		if (vol->secure_ino) {
2297			iput(vol->secure_ino);
2298			vol->secure_ino = NULL;
2299		}
2300	}
2301
2302	iput(vol->root_ino);
2303	vol->root_ino = NULL;
2304
2305	down_write(&vol->lcnbmp_lock);
2306	iput(vol->lcnbmp_ino);
2307	vol->lcnbmp_ino = NULL;
2308	up_write(&vol->lcnbmp_lock);
2309
2310	down_write(&vol->mftbmp_lock);
2311	iput(vol->mftbmp_ino);
2312	vol->mftbmp_ino = NULL;
2313	up_write(&vol->mftbmp_lock);
2314
2315#ifdef NTFS_RW
2316	if (vol->logfile_ino) {
2317		iput(vol->logfile_ino);
2318		vol->logfile_ino = NULL;
2319	}
2320	if (vol->mftmirr_ino) {
2321		/* Re-commit the mft mirror and mft just in case. */
2322		ntfs_commit_inode(vol->mftmirr_ino);
2323		ntfs_commit_inode(vol->mft_ino);
2324		iput(vol->mftmirr_ino);
2325		vol->mftmirr_ino = NULL;
2326	}
2327	/*
2328	 * If any dirty inodes are left, throw away all mft data page cache
2329	 * pages to allow a clean umount.  This should never happen any more
2330	 * due to mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2331	 * the underlying mft records are written out and cleaned.  If it does,
2332	 * happen anyway, we want to know...
2333	 */
2334	ntfs_commit_inode(vol->mft_ino);
2335	write_inode_now(vol->mft_ino, 1);
2336	if (!list_empty(&sb->s_dirty)) {
2337		const char *s1, *s2;
2338
2339		mutex_lock(&vol->mft_ino->i_mutex);
2340		truncate_inode_pages(vol->mft_ino->i_mapping, 0);
2341		mutex_unlock(&vol->mft_ino->i_mutex);
2342		write_inode_now(vol->mft_ino, 1);
2343		if (!list_empty(&sb->s_dirty)) {
2344			static const char *_s1 = "inodes";
2345			static const char *_s2 = "";
2346			s1 = _s1;
2347			s2 = _s2;
2348		} else {
2349			static const char *_s1 = "mft pages";
2350			static const char *_s2 = "They have been thrown "
2351					"away.  ";
2352			s1 = _s1;
2353			s2 = _s2;
2354		}
2355		ntfs_error(sb, "Dirty %s found at umount time.  %sYou should "
2356				"run chkdsk.  Please email "
2357				"linux-ntfs-dev@lists.sourceforge.net and say "
2358				"that you saw this message.  Thank you.", s1,
2359				s2);
2360	}
2361#endif /* NTFS_RW */
2362
2363	iput(vol->mft_ino);
2364	vol->mft_ino = NULL;
2365
2366	/* Throw away the table of attribute definitions. */
2367	vol->attrdef_size = 0;
2368	if (vol->attrdef) {
2369		ntfs_free(vol->attrdef);
2370		vol->attrdef = NULL;
2371	}
2372	vol->upcase_len = 0;
2373	/*
2374	 * Destroy the global default upcase table if necessary.  Also decrease
2375	 * the number of upcase users if we are a user.
2376	 */
2377	mutex_lock(&ntfs_lock);
2378	if (vol->upcase == default_upcase) {
2379		ntfs_nr_upcase_users--;
2380		vol->upcase = NULL;
2381	}
2382	if (!ntfs_nr_upcase_users && default_upcase) {
2383		ntfs_free(default_upcase);
2384		default_upcase = NULL;
2385	}
2386	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2387		free_compression_buffers();
2388	mutex_unlock(&ntfs_lock);
2389	if (vol->upcase) {
2390		ntfs_free(vol->upcase);
2391		vol->upcase = NULL;
2392	}
2393	if (vol->nls_map) {
2394		unload_nls(vol->nls_map);
2395		vol->nls_map = NULL;
2396	}
2397	sb->s_fs_info = NULL;
2398	kfree(vol);
2399	return;
2400}
2401
2402/**
2403 * get_nr_free_clusters - return the number of free clusters on a volume
2404 * @vol:	ntfs volume for which to obtain free cluster count
2405 *
2406 * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2407 * actually calculate the number of clusters in use instead because this
2408 * allows us to not care about partial pages as these will be just zero filled
2409 * and hence not be counted as allocated clusters.
2410 *
2411 * The only particularity is that clusters beyond the end of the logical ntfs
2412 * volume will be marked as allocated to prevent errors which means we have to
2413 * discount those at the end. This is important as the cluster bitmap always
2414 * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2415 * the logical volume and marked in use when they are not as they do not exist.
2416 *
2417 * If any pages cannot be read we assume all clusters in the erroring pages are
2418 * in use. This means we return an underestimate on errors which is better than
2419 * an overestimate.
2420 */
2421static s64 get_nr_free_clusters(ntfs_volume *vol)
2422{
2423	s64 nr_free = vol->nr_clusters;
2424	u32 *kaddr;
2425	struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2426	struct page *page;
2427	pgoff_t index, max_index;
2428
2429	ntfs_debug("Entering.");
2430	/* Serialize accesses to the cluster bitmap. */
2431	down_read(&vol->lcnbmp_lock);
2432	/*
2433	 * Convert the number of bits into bytes rounded up, then convert into
2434	 * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2435	 * full and one partial page max_index = 2.
2436	 */
2437	max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2438			PAGE_CACHE_SHIFT;
2439	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2440	ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2441			max_index, PAGE_CACHE_SIZE / 4);
2442	for (index = 0; index < max_index; index++) {
2443		unsigned int i;
2444		/*
2445		 * Read the page from page cache, getting it from backing store
2446		 * if necessary, and increment the use count.
2447		 */
2448		page = read_mapping_page(mapping, index, NULL);
2449		/* Ignore pages which errored synchronously. */
2450		if (IS_ERR(page)) {
2451			ntfs_debug("read_mapping_page() error. Skipping "
2452					"page (index 0x%lx).", index);
2453			nr_free -= PAGE_CACHE_SIZE * 8;
2454			continue;
2455		}
2456		kaddr = (u32*)kmap_atomic(page, KM_USER0);
2457		/*
2458		 * For each 4 bytes, subtract the number of set bits. If this
2459		 * is the last page and it is partial we don't really care as
2460		 * it just means we do a little extra work but it won't affect
2461		 * the result as all out of range bytes are set to zero by
2462		 * ntfs_readpage().
2463		 */
2464	  	for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2465			nr_free -= (s64)hweight32(kaddr[i]);
2466		kunmap_atomic(kaddr, KM_USER0);
2467		page_cache_release(page);
2468	}
2469	ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2470	/*
2471	 * Fixup for eventual bits outside logical ntfs volume (see function
2472	 * description above).
2473	 */
2474	if (vol->nr_clusters & 63)
2475		nr_free += 64 - (vol->nr_clusters & 63);
2476	up_read(&vol->lcnbmp_lock);
2477	/* If errors occured we may well have gone below zero, fix this. */
2478	if (nr_free < 0)
2479		nr_free = 0;
2480	ntfs_debug("Exiting.");
2481	return nr_free;
2482}
2483
2484/**
2485 * __get_nr_free_mft_records - return the number of free inodes on a volume
2486 * @vol:	ntfs volume for which to obtain free inode count
2487 * @nr_free:	number of mft records in filesystem
2488 * @max_index:	maximum number of pages containing set bits
2489 *
2490 * Calculate the number of free mft records (inodes) on the mounted NTFS
2491 * volume @vol. We actually calculate the number of mft records in use instead
2492 * because this allows us to not care about partial pages as these will be just
2493 * zero filled and hence not be counted as allocated mft record.
2494 *
2495 * If any pages cannot be read we assume all mft records in the erroring pages
2496 * are in use. This means we return an underestimate on errors which is better
2497 * than an overestimate.
2498 *
2499 * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2500 */
2501static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2502		s64 nr_free, const pgoff_t max_index)
2503{
2504	u32 *kaddr;
2505	struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2506	struct page *page;
2507	pgoff_t index;
2508
2509	ntfs_debug("Entering.");
2510	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2511	ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2512			"0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2513	for (index = 0; index < max_index; index++) {
2514		unsigned int i;
2515		/*
2516		 * Read the page from page cache, getting it from backing store
2517		 * if necessary, and increment the use count.
2518		 */
2519		page = read_mapping_page(mapping, index, NULL);
2520		/* Ignore pages which errored synchronously. */
2521		if (IS_ERR(page)) {
2522			ntfs_debug("read_mapping_page() error. Skipping "
2523					"page (index 0x%lx).", index);
2524			nr_free -= PAGE_CACHE_SIZE * 8;
2525			continue;
2526		}
2527		kaddr = (u32*)kmap_atomic(page, KM_USER0);
2528		/*
2529		 * For each 4 bytes, subtract the number of set bits. If this
2530		 * is the last page and it is partial we don't really care as
2531		 * it just means we do a little extra work but it won't affect
2532		 * the result as all out of range bytes are set to zero by
2533		 * ntfs_readpage().
2534		 */
2535	  	for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2536			nr_free -= (s64)hweight32(kaddr[i]);
2537		kunmap_atomic(kaddr, KM_USER0);
2538		page_cache_release(page);
2539	}
2540	ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2541			index - 1);
2542	/* If errors occured we may well have gone below zero, fix this. */
2543	if (nr_free < 0)
2544		nr_free = 0;
2545	ntfs_debug("Exiting.");
2546	return nr_free;
2547}
2548
2549/**
2550 * ntfs_statfs - return information about mounted NTFS volume
2551 * @dentry:	dentry from mounted volume
2552 * @sfs:	statfs structure in which to return the information
2553 *
2554 * Return information about the mounted NTFS volume @dentry in the statfs structure
2555 * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2556 * called). We interpret the values to be correct of the moment in time at
2557 * which we are called. Most values are variable otherwise and this isn't just
2558 * the free values but the totals as well. For example we can increase the
2559 * total number of file nodes if we run out and we can keep doing this until
2560 * there is no more space on the volume left at all.
2561 *
2562 * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2563 * ustat system calls.
2564 *
2565 * Return 0 on success or -errno on error.
2566 */
2567static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2568{
2569	struct super_block *sb = dentry->d_sb;
2570	s64 size;
2571	ntfs_volume *vol = NTFS_SB(sb);
2572	ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2573	pgoff_t max_index;
2574	unsigned long flags;
2575
2576	ntfs_debug("Entering.");
2577	/* Type of filesystem. */
2578	sfs->f_type   = NTFS_SB_MAGIC;
2579	/* Optimal transfer block size. */
2580	sfs->f_bsize  = PAGE_CACHE_SIZE;
2581	/*
2582	 * Total data blocks in filesystem in units of f_bsize and since
2583	 * inodes are also stored in data blocs ($MFT is a file) this is just
2584	 * the total clusters.
2585	 */
2586	sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2587				PAGE_CACHE_SHIFT;
2588	/* Free data blocks in filesystem in units of f_bsize. */
2589	size	      = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2590				PAGE_CACHE_SHIFT;
2591	if (size < 0LL)
2592		size = 0LL;
2593	/* Free blocks avail to non-superuser, same as above on NTFS. */
2594	sfs->f_bavail = sfs->f_bfree = size;
2595	/* Serialize accesses to the inode bitmap. */
2596	down_read(&vol->mftbmp_lock);
2597	read_lock_irqsave(&mft_ni->size_lock, flags);
2598	size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2599	/*
2600	 * Convert the maximum number of set bits into bytes rounded up, then
2601	 * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2602	 * have one full and one partial page max_index = 2.
2603	 */
2604	max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2605			+ 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2606	read_unlock_irqrestore(&mft_ni->size_lock, flags);
2607	/* Number of inodes in filesystem (at this point in time). */
2608	sfs->f_files = size;
2609	/* Free inodes in fs (based on current total count). */
2610	sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2611	up_read(&vol->mftbmp_lock);
2612	/*
2613	 * File system id. This is extremely *nix flavour dependent and even
2614	 * within Linux itself all fs do their own thing. I interpret this to
2615	 * mean a unique id associated with the mounted fs and not the id
2616	 * associated with the filesystem driver, the latter is already given
2617	 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2618	 * volume serial number splitting it into two 32-bit parts. We enter
2619	 * the least significant 32-bits in f_fsid[0] and the most significant
2620	 * 32-bits in f_fsid[1].
2621	 */
2622	sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2623	sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2624	/* Maximum length of filenames. */
2625	sfs->f_namelen	   = NTFS_MAX_NAME_LEN;
2626	return 0;
2627}
2628
2629/**
2630 * The complete super operations.
2631 */
2632static const struct super_operations ntfs_sops = {
2633	.alloc_inode	= ntfs_alloc_big_inode,	  /* VFS: Allocate new inode. */
2634	.destroy_inode	= ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2635#ifdef NTFS_RW
2636	//.dirty_inode	= NULL,			/* VFS: Called from
2637	//					   __mark_inode_dirty(). */
2638	.write_inode	= ntfs_write_inode,	/* VFS: Write dirty inode to
2639						   disk. */
2640	//.drop_inode	= NULL,			/* VFS: Called just after the
2641	//					   inode reference count has
2642	//					   been decreased to zero.
2643	//					   NOTE: The inode lock is
2644	//					   held. See fs/inode.c::
2645	//					   generic_drop_inode(). */
2646	//.delete_inode	= NULL,			/* VFS: Delete inode from disk.
2647	//					   Called when i_count becomes
2648	//					   0 and i_nlink is also 0. */
2649	//.write_super	= NULL,			/* Flush dirty super block to
2650	//					   disk. */
2651	//.sync_fs	= NULL,			/* ? */
2652	//.write_super_lockfs	= NULL,		/* ? */
2653	//.unlockfs	= NULL,			/* ? */
2654#endif /* NTFS_RW */
2655	.put_super	= ntfs_put_super,	/* Syscall: umount. */
2656	.statfs		= ntfs_statfs,		/* Syscall: statfs */
2657	.remount_fs	= ntfs_remount,		/* Syscall: mount -o remount. */
2658	.clear_inode	= ntfs_clear_big_inode,	/* VFS: Called when an inode is
2659						   removed from memory. */
2660	//.umount_begin	= NULL,			/* Forced umount. */
2661	.show_options	= ntfs_show_options,	/* Show mount options in
2662						   proc. */
2663};
2664
2665/**
2666 * ntfs_fill_super - mount an ntfs filesystem
2667 * @sb:		super block of ntfs filesystem to mount
2668 * @opt:	string containing the mount options
2669 * @silent:	silence error output
2670 *
2671 * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2672 * with the mount otions in @data with the NTFS filesystem.
2673 *
2674 * If @silent is true, remain silent even if errors are detected. This is used
2675 * during bootup, when the kernel tries to mount the root filesystem with all
2676 * registered filesystems one after the other until one succeeds. This implies
2677 * that all filesystems except the correct one will quite correctly and
2678 * expectedly return an error, but nobody wants to see error messages when in
2679 * fact this is what is supposed to happen.
2680 *
2681 * NOTE: @sb->s_flags contains the mount options flags.
2682 */
2683static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2684{
2685	ntfs_volume *vol;
2686	struct buffer_head *bh;
2687	struct inode *tmp_ino;
2688	int blocksize, result;
2689
2690	/*
2691	 * We do a pretty difficult piece of bootstrap by reading the
2692	 * MFT (and other metadata) from disk into memory. We'll only
2693	 * release this metadata during umount, so the locking patterns
2694	 * observed during bootstrap do not count. So turn off the
2695	 * observation of locking patterns (strictly for this context
2696	 * only) while mounting NTFS. [The validator is still active
2697	 * otherwise, even for this context: it will for example record
2698	 * lock class registrations.]
2699	 */
2700	lockdep_off();
2701	ntfs_debug("Entering.");
2702#ifndef NTFS_RW
2703	sb->s_flags |= MS_RDONLY;
2704#endif /* ! NTFS_RW */
2705	/* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2706	sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2707	vol = NTFS_SB(sb);
2708	if (!vol) {
2709		if (!silent)
2710			ntfs_error(sb, "Allocation of NTFS volume structure "
2711					"failed. Aborting mount...");
2712		lockdep_on();
2713		return -ENOMEM;
2714	}
2715	/* Initialize ntfs_volume structure. */
2716	*vol = (ntfs_volume) {
2717		.sb = sb,
2718		/*
2719		 * Default is group and other don't have any access to files or
2720		 * directories while owner has full access. Further, files by
2721		 * default are not executable but directories are of course
2722		 * browseable.
2723		 */
2724		.fmask = 0177,
2725		.dmask = 0077,
2726	};
2727	init_rwsem(&vol->mftbmp_lock);
2728	init_rwsem(&vol->lcnbmp_lock);
2729
2730	unlock_kernel();
2731
2732	/* By default, enable sparse support. */
2733	NVolSetSparseEnabled(vol);
2734
2735	/* Important to get the mount options dealt with now. */
2736	if (!parse_options(vol, (char*)opt))
2737		goto err_out_now;
2738
2739	/* We support sector sizes up to the PAGE_CACHE_SIZE. */
2740	if (bdev_hardsect_size(sb->s_bdev) > PAGE_CACHE_SIZE) {
2741		if (!silent)
2742			ntfs_error(sb, "Device has unsupported sector size "
2743					"(%i).  The maximum supported sector "
2744					"size on this architecture is %lu "
2745					"bytes.",
2746					bdev_hardsect_size(sb->s_bdev),
2747					PAGE_CACHE_SIZE);
2748		goto err_out_now;
2749	}
2750	/*
2751	 * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2752	 * sector size, whichever is bigger.
2753	 */
2754	blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2755	if (blocksize < NTFS_BLOCK_SIZE) {
2756		if (!silent)
2757			ntfs_error(sb, "Unable to set device block size.");
2758		goto err_out_now;
2759	}
2760	BUG_ON(blocksize != sb->s_blocksize);
2761	ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2762			blocksize, sb->s_blocksize_bits);
2763	/* Determine the size of the device in units of block_size bytes. */
2764	if (!i_size_read(sb->s_bdev->bd_inode)) {
2765		if (!silent)
2766			ntfs_error(sb, "Unable to determine device size.");
2767		goto err_out_now;
2768	}
2769	vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2770			sb->s_blocksize_bits;
2771	/* Read the boot sector and return unlocked buffer head to it. */
2772	if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2773		if (!silent)
2774			ntfs_error(sb, "Not an NTFS volume.");
2775		goto err_out_now;
2776	}
2777	/*
2778	 * Extract the data from the boot sector and setup the ntfs volume
2779	 * using it.
2780	 */
2781	result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2782	brelse(bh);
2783	if (!result) {
2784		if (!silent)
2785			ntfs_error(sb, "Unsupported NTFS filesystem.");
2786		goto err_out_now;
2787	}
2788	/*
2789	 * If the boot sector indicates a sector size bigger than the current
2790	 * device block size, switch the device block size to the sector size.
2791	 * TODO: It may be possible to support this case even when the set
2792	 * below fails, we would just be breaking up the i/o for each sector
2793	 * into multiple blocks for i/o purposes but otherwise it should just
2794	 * work.  However it is safer to leave disabled until someone hits this
2795	 * error message and then we can get them to try it without the setting
2796	 * so we know for sure that it works.
2797	 */
2798	if (vol->sector_size > blocksize) {
2799		blocksize = sb_set_blocksize(sb, vol->sector_size);
2800		if (blocksize != vol->sector_size) {
2801			if (!silent)
2802				ntfs_error(sb, "Unable to set device block "
2803						"size to sector size (%i).",
2804						vol->sector_size);
2805			goto err_out_now;
2806		}
2807		BUG_ON(blocksize != sb->s_blocksize);
2808		vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2809				sb->s_blocksize_bits;
2810		ntfs_debug("Changed device block size to %i bytes (block size "
2811				"bits %i) to match volume sector size.",
2812				blocksize, sb->s_blocksize_bits);
2813	}
2814	/* Initialize the cluster and mft allocators. */
2815	ntfs_setup_allocators(vol);
2816	/* Setup remaining fields in the super block. */
2817	sb->s_magic = NTFS_SB_MAGIC;
2818	/*
2819	 * Ntfs allows 63 bits for the file size, i.e. correct would be:
2820	 *	sb->s_maxbytes = ~0ULL >> 1;
2821	 * But the kernel uses a long as the page cache page index which on
2822	 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2823	 * defined to the maximum the page cache page index can cope with
2824	 * without overflowing the index or to 2^63 - 1, whichever is smaller.
2825	 */
2826	sb->s_maxbytes = MAX_LFS_FILESIZE;
2827	/* Ntfs measures time in 100ns intervals. */
2828	sb->s_time_gran = 100;
2829	/*
2830	 * Now load the metadata required for the page cache and our address
2831	 * space operations to function. We do this by setting up a specialised
2832	 * read_inode method and then just calling the normal iget() to obtain
2833	 * the inode for $MFT which is sufficient to allow our normal inode
2834	 * operations and associated address space operations to function.
2835	 */
2836	sb->s_op = &ntfs_sops;
2837	tmp_ino = new_inode(sb);
2838	if (!tmp_ino) {
2839		if (!silent)
2840			ntfs_error(sb, "Failed to load essential metadata.");
2841		goto err_out_now;
2842	}
2843	tmp_ino->i_ino = FILE_MFT;
2844	insert_inode_hash(tmp_ino);
2845	if (ntfs_read_inode_mount(tmp_ino) < 0) {
2846		if (!silent)
2847			ntfs_error(sb, "Failed to load essential metadata.");
2848		goto iput_tmp_ino_err_out_now;
2849	}
2850	mutex_lock(&ntfs_lock);
2851	/*
2852	 * The current mount is a compression user if the cluster size is
2853	 * less than or equal 4kiB.
2854	 */
2855	if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2856		result = allocate_compression_buffers();
2857		if (result) {
2858			ntfs_error(NULL, "Failed to allocate buffers "
2859					"for compression engine.");
2860			ntfs_nr_compression_users--;
2861			mutex_unlock(&ntfs_lock);
2862			goto iput_tmp_ino_err_out_now;
2863		}
2864	}
2865	/*
2866	 * Generate the global default upcase table if necessary.  Also
2867	 * temporarily increment the number of upcase users to avoid race
2868	 * conditions with concurrent (u)mounts.
2869	 */
2870	if (!default_upcase)
2871		default_upcase = generate_default_upcase();
2872	ntfs_nr_upcase_users++;
2873	mutex_unlock(&ntfs_lock);
2874	/*
2875	 * From now on, ignore @silent parameter. If we fail below this line,
2876	 * it will be due to a corrupt fs or a system error, so we report it.
2877	 */
2878	/*
2879	 * Open the system files with normal access functions and complete
2880	 * setting up the ntfs super block.
2881	 */
2882	if (!load_system_files(vol)) {
2883		ntfs_error(sb, "Failed to load system files.");
2884		goto unl_upcase_iput_tmp_ino_err_out_now;
2885	}
2886	if ((sb->s_root = d_alloc_root(vol->root_ino))) {
2887		/* We increment i_count simulating an ntfs_iget(). */
2888		atomic_inc(&vol->root_ino->i_count);
2889		ntfs_debug("Exiting, status successful.");
2890		/* Release the default upcase if it has no users. */
2891		mutex_lock(&ntfs_lock);
2892		if (!--ntfs_nr_upcase_users && default_upcase) {
2893			ntfs_free(default_upcase);
2894			default_upcase = NULL;
2895		}
2896		mutex_unlock(&ntfs_lock);
2897		sb->s_export_op = &ntfs_export_ops;
2898		lock_kernel();
2899		lockdep_on();
2900		return 0;
2901	}
2902	ntfs_error(sb, "Failed to allocate root directory.");
2903	/* Clean up after the successful load_system_files() call from above. */
2904	// TODO: Use ntfs_put_super() instead of repeating all this code...
2905	// 	  -ENOMEM.
2906	iput(vol->vol_ino);
2907	vol->vol_ino = NULL;
2908	/* NTFS 3.0+ specific clean up. */
2909	if (vol->major_ver >= 3) {
2910#ifdef NTFS_RW
2911		if (vol->usnjrnl_j_ino) {
2912			iput(vol->usnjrnl_j_ino);
2913			vol->usnjrnl_j_ino = NULL;
2914		}
2915		if (vol->usnjrnl_max_ino) {
2916			iput(vol->usnjrnl_max_ino);
2917			vol->usnjrnl_max_ino = NULL;
2918		}
2919		if (vol->usnjrnl_ino) {
2920			iput(vol->usnjrnl_ino);
2921			vol->usnjrnl_ino = NULL;
2922		}
2923		if (vol->quota_q_ino) {
2924			iput(vol->quota_q_ino);
2925			vol->quota_q_ino = NULL;
2926		}
2927		if (vol->quota_ino) {
2928			iput(vol->quota_ino);
2929			vol->quota_ino = NULL;
2930		}
2931#endif /* NTFS_RW */
2932		if (vol->extend_ino) {
2933			iput(vol->extend_ino);
2934			vol->extend_ino = NULL;
2935		}
2936		if (vol->secure_ino) {
2937			iput(vol->secure_ino);
2938			vol->secure_ino = NULL;
2939		}
2940	}
2941	iput(vol->root_ino);
2942	vol->root_ino = NULL;
2943	iput(vol->lcnbmp_ino);
2944	vol->lcnbmp_ino = NULL;
2945	iput(vol->mftbmp_ino);
2946	vol->mftbmp_ino = NULL;
2947#ifdef NTFS_RW
2948	if (vol->logfile_ino) {
2949		iput(vol->logfile_ino);
2950		vol->logfile_ino = NULL;
2951	}
2952	if (vol->mftmirr_ino) {
2953		iput(vol->mftmirr_ino);
2954		vol->mftmirr_ino = NULL;
2955	}
2956#endif /* NTFS_RW */
2957	/* Throw away the table of attribute definitions. */
2958	vol->attrdef_size = 0;
2959	if (vol->attrdef) {
2960		ntfs_free(vol->attrdef);
2961		vol->attrdef = NULL;
2962	}
2963	vol->upcase_len = 0;
2964	mutex_lock(&ntfs_lock);
2965	if (vol->upcase == default_upcase) {
2966		ntfs_nr_upcase_users--;
2967		vol->upcase = NULL;
2968	}
2969	mutex_unlock(&ntfs_lock);
2970	if (vol->upcase) {
2971		ntfs_free(vol->upcase);
2972		vol->upcase = NULL;
2973	}
2974	if (vol->nls_map) {
2975		unload_nls(vol->nls_map);
2976		vol->nls_map = NULL;
2977	}
2978	/* Error exit code path. */
2979unl_upcase_iput_tmp_ino_err_out_now:
2980	/*
2981	 * Decrease the number of upcase users and destroy the global default
2982	 * upcase table if necessary.
2983	 */
2984	mutex_lock(&ntfs_lock);
2985	if (!--ntfs_nr_upcase_users && default_upcase) {
2986		ntfs_free(default_upcase);
2987		default_upcase = NULL;
2988	}
2989	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2990		free_compression_buffers();
2991	mutex_unlock(&ntfs_lock);
2992iput_tmp_ino_err_out_now:
2993	iput(tmp_ino);
2994	if (vol->mft_ino && vol->mft_ino != tmp_ino)
2995		iput(vol->mft_ino);
2996	vol->mft_ino = NULL;
2997	if (invalidate_inodes(sb)) {
2998		ntfs_error(sb, "Busy inodes left. This is most likely a NTFS "
2999				"driver bug.");
3000		/* Copied from fs/super.c. I just love this message. (-; */
3001		printk("NTFS: Busy inodes after umount. Self-destruct in 5 "
3002				"seconds.  Have a nice day...\n");
3003	}
3004	/* Errors at this stage are irrelevant. */
3005err_out_now:
3006	lock_kernel();
3007	sb->s_fs_info = NULL;
3008	kfree(vol);
3009	ntfs_debug("Failed, returning -EINVAL.");
3010	lockdep_on();
3011	return -EINVAL;
3012}
3013
3014/*
3015 * This is a slab cache to optimize allocations and deallocations of Unicode
3016 * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3017 * (255) Unicode characters + a terminating NULL Unicode character.
3018 */
3019struct kmem_cache *ntfs_name_cache;
3020
3021/* Slab caches for efficient allocation/deallocation of inodes. */
3022struct kmem_cache *ntfs_inode_cache;
3023struct kmem_cache *ntfs_big_inode_cache;
3024
3025/* Init once constructor for the inode slab cache. */
3026static void ntfs_big_inode_init_once(void *foo, struct kmem_cache *cachep,
3027		unsigned long flags)
3028{
3029	ntfs_inode *ni = (ntfs_inode *)foo;
3030
3031	inode_init_once(VFS_I(ni));
3032}
3033
3034/*
3035 * Slab caches to optimize allocations and deallocations of attribute search
3036 * contexts and index contexts, respectively.
3037 */
3038struct kmem_cache *ntfs_attr_ctx_cache;
3039struct kmem_cache *ntfs_index_ctx_cache;
3040
3041/* Driver wide mutex. */
3042DEFINE_MUTEX(ntfs_lock);
3043
3044static int ntfs_get_sb(struct file_system_type *fs_type,
3045	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
3046{
3047	return get_sb_bdev(fs_type, flags, dev_name, data, ntfs_fill_super,
3048			   mnt);
3049}
3050
3051static struct file_system_type ntfs_fs_type = {
3052	.owner		= THIS_MODULE,
3053	.name		= "ntfs",
3054	.get_sb		= ntfs_get_sb,
3055	.kill_sb	= kill_block_super,
3056	.fs_flags	= FS_REQUIRES_DEV,
3057};
3058
3059/* Stable names for the slab caches. */
3060static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3061static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3062static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3063static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3064static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3065
3066static int __init init_ntfs_fs(void)
3067{
3068	int err = 0;
3069
3070	/* This may be ugly but it results in pretty output so who cares. (-8 */
3071	printk(KERN_INFO "NTFS driver " NTFS_VERSION " [Flags: R/"
3072#ifdef NTFS_RW
3073			"W"
3074#else
3075			"O"
3076#endif
3077#ifdef DEBUG
3078			" DEBUG"
3079#endif
3080#ifdef MODULE
3081			" MODULE"
3082#endif
3083			"].\n");
3084
3085	ntfs_debug("Debug messages are enabled.");
3086
3087	ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3088			sizeof(ntfs_index_context), 0 /* offset */,
3089			SLAB_HWCACHE_ALIGN, NULL /* ctor */, NULL /* dtor */);
3090	if (!ntfs_index_ctx_cache) {
3091		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3092				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 */, NULL /* dtor */);
3098	if (!ntfs_attr_ctx_cache) {
3099		printk(KERN_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, NULL);
3107	if (!ntfs_name_cache) {
3108		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3109				ntfs_name_cache_name);
3110		goto name_err_out;
3111	}
3112
3113	ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3114			sizeof(ntfs_inode), 0,
3115			SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL, NULL);
3116	if (!ntfs_inode_cache) {
3117		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3118				ntfs_inode_cache_name);
3119		goto inode_err_out;
3120	}
3121
3122	ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3123			sizeof(big_ntfs_inode), 0,
3124			SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
3125			ntfs_big_inode_init_once, NULL);
3126	if (!ntfs_big_inode_cache) {
3127		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3128				ntfs_big_inode_cache_name);
3129		goto big_inode_err_out;
3130	}
3131
3132	/* Register the ntfs sysctls. */
3133	err = ntfs_sysctl(1);
3134	if (err) {
3135		printk(KERN_CRIT "NTFS: Failed to register NTFS sysctls!\n");
3136		goto sysctl_err_out;
3137	}
3138
3139	err = register_filesystem(&ntfs_fs_type);
3140	if (!err) {
3141		ntfs_debug("NTFS driver registered successfully.");
3142		return 0; /* Success! */
3143	}
3144	printk(KERN_CRIT "NTFS: Failed to register NTFS filesystem driver!\n");
3145
3146sysctl_err_out:
3147	kmem_cache_destroy(ntfs_big_inode_cache);
3148big_inode_err_out:
3149	kmem_cache_destroy(ntfs_inode_cache);
3150inode_err_out:
3151	kmem_cache_destroy(ntfs_name_cache);
3152name_err_out:
3153	kmem_cache_destroy(ntfs_attr_ctx_cache);
3154actx_err_out:
3155	kmem_cache_destroy(ntfs_index_ctx_cache);
3156ictx_err_out:
3157	if (!err) {
3158		printk(KERN_CRIT "NTFS: Aborting NTFS filesystem driver "
3159				"registration...\n");
3160		err = -ENOMEM;
3161	}
3162	return err;
3163}
3164
3165static void __exit exit_ntfs_fs(void)
3166{
3167	ntfs_debug("Unregistering NTFS driver.");
3168
3169	unregister_filesystem(&ntfs_fs_type);
3170	kmem_cache_destroy(ntfs_big_inode_cache);
3171	kmem_cache_destroy(ntfs_inode_cache);
3172	kmem_cache_destroy(ntfs_name_cache);
3173	kmem_cache_destroy(ntfs_attr_ctx_cache);
3174	kmem_cache_destroy(ntfs_index_ctx_cache);
3175	/* Unregister the ntfs sysctls. */
3176	ntfs_sysctl(0);
3177}
3178
3179MODULE_AUTHOR("Anton Altaparmakov <aia21@cantab.net>");
3180MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2007 Anton Altaparmakov");
3181MODULE_VERSION(NTFS_VERSION);
3182MODULE_LICENSE("GPL");
3183#ifdef DEBUG
3184module_param(debug_msgs, bool, 0);
3185MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3186#endif
3187
3188module_init(init_ntfs_fs)
3189module_exit(exit_ntfs_fs)
3190