xref: /openbmc/linux/fs/ntfs/super.c (revision 5d46770f)
1 /*
2  * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
3  *
4  * Copyright (c) 2001-2005 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/string.h>
26 #include <linux/spinlock.h>
27 #include <linux/blkdev.h>	/* For bdev_hardsect_size(). */
28 #include <linux/backing-dev.h>
29 #include <linux/buffer_head.h>
30 #include <linux/vfs.h>
31 #include <linux/moduleparam.h>
32 #include <linux/smp_lock.h>
33 
34 #include "sysctl.h"
35 #include "logfile.h"
36 #include "quota.h"
37 #include "usnjrnl.h"
38 #include "dir.h"
39 #include "debug.h"
40 #include "index.h"
41 #include "aops.h"
42 #include "layout.h"
43 #include "malloc.h"
44 #include "ntfs.h"
45 
46 /* Number of mounted filesystems which have compression enabled. */
47 static unsigned long ntfs_nr_compression_users;
48 
49 /* A global default upcase table and a corresponding reference count. */
50 static ntfschar *default_upcase = NULL;
51 static unsigned long ntfs_nr_upcase_users = 0;
52 
53 /* Error constants/strings used in inode.c::ntfs_show_options(). */
54 typedef enum {
55 	/* One of these must be present, default is ON_ERRORS_CONTINUE. */
56 	ON_ERRORS_PANIC			= 0x01,
57 	ON_ERRORS_REMOUNT_RO		= 0x02,
58 	ON_ERRORS_CONTINUE		= 0x04,
59 	/* Optional, can be combined with any of the above. */
60 	ON_ERRORS_RECOVER		= 0x10,
61 } ON_ERRORS_ACTIONS;
62 
63 const option_t on_errors_arr[] = {
64 	{ ON_ERRORS_PANIC,	"panic" },
65 	{ ON_ERRORS_REMOUNT_RO,	"remount-ro", },
66 	{ ON_ERRORS_CONTINUE,	"continue", },
67 	{ ON_ERRORS_RECOVER,	"recover" },
68 	{ 0,			NULL }
69 };
70 
71 /**
72  * simple_getbool -
73  *
74  * Copied from old ntfs driver (which copied from vfat driver).
75  */
76 static int simple_getbool(char *s, BOOL *setval)
77 {
78 	if (s) {
79 		if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
80 			*setval = TRUE;
81 		else if (!strcmp(s, "0") || !strcmp(s, "no") ||
82 							!strcmp(s, "false"))
83 			*setval = FALSE;
84 		else
85 			return 0;
86 	} else
87 		*setval = TRUE;
88 	return 1;
89 }
90 
91 /**
92  * parse_options - parse the (re)mount options
93  * @vol:	ntfs volume
94  * @opt:	string containing the (re)mount options
95  *
96  * Parse the recognized options in @opt for the ntfs volume described by @vol.
97  */
98 static BOOL parse_options(ntfs_volume *vol, char *opt)
99 {
100 	char *p, *v, *ov;
101 	static char *utf8 = "utf8";
102 	int errors = 0, sloppy = 0;
103 	uid_t uid = (uid_t)-1;
104 	gid_t gid = (gid_t)-1;
105 	mode_t fmask = (mode_t)-1, dmask = (mode_t)-1;
106 	int mft_zone_multiplier = -1, on_errors = -1;
107 	int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
108 	struct nls_table *nls_map = NULL, *old_nls;
109 
110 	/* I am lazy... (-8 */
111 #define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value)	\
112 	if (!strcmp(p, option)) {					\
113 		if (!v || !*v)						\
114 			variable = default_value;			\
115 		else {							\
116 			variable = simple_strtoul(ov = v, &v, 0);	\
117 			if (*v)						\
118 				goto needs_val;				\
119 		}							\
120 	}
121 #define NTFS_GETOPT(option, variable)					\
122 	if (!strcmp(p, option)) {					\
123 		if (!v || !*v)						\
124 			goto needs_arg;					\
125 		variable = simple_strtoul(ov = v, &v, 0);		\
126 		if (*v)							\
127 			goto needs_val;					\
128 	}
129 #define NTFS_GETOPT_OCTAL(option, variable)				\
130 	if (!strcmp(p, option)) {					\
131 		if (!v || !*v)						\
132 			goto needs_arg;					\
133 		variable = simple_strtoul(ov = v, &v, 8);		\
134 		if (*v)							\
135 			goto needs_val;					\
136 	}
137 #define NTFS_GETOPT_BOOL(option, variable)				\
138 	if (!strcmp(p, option)) {					\
139 		BOOL val;						\
140 		if (!simple_getbool(v, &val))				\
141 			goto needs_bool;				\
142 		variable = val;						\
143 	}
144 #define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array)		\
145 	if (!strcmp(p, option)) {					\
146 		int _i;							\
147 		if (!v || !*v)						\
148 			goto needs_arg;					\
149 		ov = v;							\
150 		if (variable == -1)					\
151 			variable = 0;					\
152 		for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
153 			if (!strcmp(opt_array[_i].str, v)) {		\
154 				variable |= opt_array[_i].val;		\
155 				break;					\
156 			}						\
157 		if (!opt_array[_i].str || !*opt_array[_i].str)		\
158 			goto needs_val;					\
159 	}
160 	if (!opt || !*opt)
161 		goto no_mount_options;
162 	ntfs_debug("Entering with mount options string: %s", opt);
163 	while ((p = strsep(&opt, ","))) {
164 		if ((v = strchr(p, '=')))
165 			*v++ = 0;
166 		NTFS_GETOPT("uid", uid)
167 		else NTFS_GETOPT("gid", gid)
168 		else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
169 		else NTFS_GETOPT_OCTAL("fmask", fmask)
170 		else NTFS_GETOPT_OCTAL("dmask", dmask)
171 		else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
172 		else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, TRUE)
173 		else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
174 		else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
175 		else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
176 		else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
177 				on_errors_arr)
178 		else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
179 			ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
180 					p);
181 		else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
182 			if (!strcmp(p, "iocharset"))
183 				ntfs_warning(vol->sb, "Option iocharset is "
184 						"deprecated. Please use "
185 						"option nls=<charsetname> in "
186 						"the future.");
187 			if (!v || !*v)
188 				goto needs_arg;
189 use_utf8:
190 			old_nls = nls_map;
191 			nls_map = load_nls(v);
192 			if (!nls_map) {
193 				if (!old_nls) {
194 					ntfs_error(vol->sb, "NLS character set "
195 							"%s not found.", v);
196 					return FALSE;
197 				}
198 				ntfs_error(vol->sb, "NLS character set %s not "
199 						"found. Using previous one %s.",
200 						v, old_nls->charset);
201 				nls_map = old_nls;
202 			} else /* nls_map */ {
203 				if (old_nls)
204 					unload_nls(old_nls);
205 			}
206 		} else if (!strcmp(p, "utf8")) {
207 			BOOL val = FALSE;
208 			ntfs_warning(vol->sb, "Option utf8 is no longer "
209 				   "supported, using option nls=utf8. Please "
210 				   "use option nls=utf8 in the future and "
211 				   "make sure utf8 is compiled either as a "
212 				   "module or into the kernel.");
213 			if (!v || !*v)
214 				val = TRUE;
215 			else if (!simple_getbool(v, &val))
216 				goto needs_bool;
217 			if (val) {
218 				v = utf8;
219 				goto use_utf8;
220 			}
221 		} else {
222 			ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
223 			if (errors < INT_MAX)
224 				errors++;
225 		}
226 #undef NTFS_GETOPT_OPTIONS_ARRAY
227 #undef NTFS_GETOPT_BOOL
228 #undef NTFS_GETOPT
229 #undef NTFS_GETOPT_WITH_DEFAULT
230 	}
231 no_mount_options:
232 	if (errors && !sloppy)
233 		return FALSE;
234 	if (sloppy)
235 		ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
236 				"unrecognized mount option(s) and continuing.");
237 	/* Keep this first! */
238 	if (on_errors != -1) {
239 		if (!on_errors) {
240 			ntfs_error(vol->sb, "Invalid errors option argument "
241 					"or bug in options parser.");
242 			return FALSE;
243 		}
244 	}
245 	if (nls_map) {
246 		if (vol->nls_map && vol->nls_map != nls_map) {
247 			ntfs_error(vol->sb, "Cannot change NLS character set "
248 					"on remount.");
249 			return FALSE;
250 		} /* else (!vol->nls_map) */
251 		ntfs_debug("Using NLS character set %s.", nls_map->charset);
252 		vol->nls_map = nls_map;
253 	} else /* (!nls_map) */ {
254 		if (!vol->nls_map) {
255 			vol->nls_map = load_nls_default();
256 			if (!vol->nls_map) {
257 				ntfs_error(vol->sb, "Failed to load default "
258 						"NLS character set.");
259 				return FALSE;
260 			}
261 			ntfs_debug("Using default NLS character set (%s).",
262 					vol->nls_map->charset);
263 		}
264 	}
265 	if (mft_zone_multiplier != -1) {
266 		if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
267 				mft_zone_multiplier) {
268 			ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
269 					"on remount.");
270 			return FALSE;
271 		}
272 		if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
273 			ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
274 					"Using default value, i.e. 1.");
275 			mft_zone_multiplier = 1;
276 		}
277 		vol->mft_zone_multiplier = mft_zone_multiplier;
278 	}
279 	if (!vol->mft_zone_multiplier)
280 		vol->mft_zone_multiplier = 1;
281 	if (on_errors != -1)
282 		vol->on_errors = on_errors;
283 	if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
284 		vol->on_errors |= ON_ERRORS_CONTINUE;
285 	if (uid != (uid_t)-1)
286 		vol->uid = uid;
287 	if (gid != (gid_t)-1)
288 		vol->gid = gid;
289 	if (fmask != (mode_t)-1)
290 		vol->fmask = fmask;
291 	if (dmask != (mode_t)-1)
292 		vol->dmask = dmask;
293 	if (show_sys_files != -1) {
294 		if (show_sys_files)
295 			NVolSetShowSystemFiles(vol);
296 		else
297 			NVolClearShowSystemFiles(vol);
298 	}
299 	if (case_sensitive != -1) {
300 		if (case_sensitive)
301 			NVolSetCaseSensitive(vol);
302 		else
303 			NVolClearCaseSensitive(vol);
304 	}
305 	if (disable_sparse != -1) {
306 		if (disable_sparse)
307 			NVolClearSparseEnabled(vol);
308 		else {
309 			if (!NVolSparseEnabled(vol) &&
310 					vol->major_ver && vol->major_ver < 3)
311 				ntfs_warning(vol->sb, "Not enabling sparse "
312 						"support due to NTFS volume "
313 						"version %i.%i (need at least "
314 						"version 3.0).", vol->major_ver,
315 						vol->minor_ver);
316 			else
317 				NVolSetSparseEnabled(vol);
318 		}
319 	}
320 	return TRUE;
321 needs_arg:
322 	ntfs_error(vol->sb, "The %s option requires an argument.", p);
323 	return FALSE;
324 needs_bool:
325 	ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
326 	return FALSE;
327 needs_val:
328 	ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
329 	return FALSE;
330 }
331 
332 #ifdef NTFS_RW
333 
334 /**
335  * ntfs_write_volume_flags - write new flags to the volume information flags
336  * @vol:	ntfs volume on which to modify the flags
337  * @flags:	new flags value for the volume information flags
338  *
339  * Internal function.  You probably want to use ntfs_{set,clear}_volume_flags()
340  * instead (see below).
341  *
342  * Replace the volume information flags on the volume @vol with the value
343  * supplied in @flags.  Note, this overwrites the volume information flags, so
344  * make sure to combine the flags you want to modify with the old flags and use
345  * the result when calling ntfs_write_volume_flags().
346  *
347  * Return 0 on success and -errno on error.
348  */
349 static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
350 {
351 	ntfs_inode *ni = NTFS_I(vol->vol_ino);
352 	MFT_RECORD *m;
353 	VOLUME_INFORMATION *vi;
354 	ntfs_attr_search_ctx *ctx;
355 	int err;
356 
357 	ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
358 			le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
359 	if (vol->vol_flags == flags)
360 		goto done;
361 	BUG_ON(!ni);
362 	m = map_mft_record(ni);
363 	if (IS_ERR(m)) {
364 		err = PTR_ERR(m);
365 		goto err_out;
366 	}
367 	ctx = ntfs_attr_get_search_ctx(ni, m);
368 	if (!ctx) {
369 		err = -ENOMEM;
370 		goto put_unm_err_out;
371 	}
372 	err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
373 			ctx);
374 	if (err)
375 		goto put_unm_err_out;
376 	vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
377 			le16_to_cpu(ctx->attr->data.resident.value_offset));
378 	vol->vol_flags = vi->flags = flags;
379 	flush_dcache_mft_record_page(ctx->ntfs_ino);
380 	mark_mft_record_dirty(ctx->ntfs_ino);
381 	ntfs_attr_put_search_ctx(ctx);
382 	unmap_mft_record(ni);
383 done:
384 	ntfs_debug("Done.");
385 	return 0;
386 put_unm_err_out:
387 	if (ctx)
388 		ntfs_attr_put_search_ctx(ctx);
389 	unmap_mft_record(ni);
390 err_out:
391 	ntfs_error(vol->sb, "Failed with error code %i.", -err);
392 	return err;
393 }
394 
395 /**
396  * ntfs_set_volume_flags - set bits in the volume information flags
397  * @vol:	ntfs volume on which to modify the flags
398  * @flags:	flags to set on the volume
399  *
400  * Set the bits in @flags in the volume information flags on the volume @vol.
401  *
402  * Return 0 on success and -errno on error.
403  */
404 static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
405 {
406 	flags &= VOLUME_FLAGS_MASK;
407 	return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
408 }
409 
410 /**
411  * ntfs_clear_volume_flags - clear bits in the volume information flags
412  * @vol:	ntfs volume on which to modify the flags
413  * @flags:	flags to clear on the volume
414  *
415  * Clear the bits in @flags in the volume information flags on the volume @vol.
416  *
417  * Return 0 on success and -errno on error.
418  */
419 static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
420 {
421 	flags &= VOLUME_FLAGS_MASK;
422 	flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
423 	return ntfs_write_volume_flags(vol, flags);
424 }
425 
426 #endif /* NTFS_RW */
427 
428 /**
429  * ntfs_remount - change the mount options of a mounted ntfs filesystem
430  * @sb:		superblock of mounted ntfs filesystem
431  * @flags:	remount flags
432  * @opt:	remount options string
433  *
434  * Change the mount options of an already mounted ntfs filesystem.
435  *
436  * NOTE:  The VFS sets the @sb->s_flags remount flags to @flags after
437  * ntfs_remount() returns successfully (i.e. returns 0).  Otherwise,
438  * @sb->s_flags are not changed.
439  */
440 static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
441 {
442 	ntfs_volume *vol = NTFS_SB(sb);
443 
444 	ntfs_debug("Entering with remount options string: %s", opt);
445 #ifndef NTFS_RW
446 	/* For read-only compiled driver, enforce all read-only flags. */
447 	*flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
448 #else /* NTFS_RW */
449 	/*
450 	 * For the read-write compiled driver, if we are remounting read-write,
451 	 * make sure there are no volume errors and that no unsupported volume
452 	 * flags are set.  Also, empty the logfile journal as it would become
453 	 * stale as soon as something is written to the volume and mark the
454 	 * volume dirty so that chkdsk is run if the volume is not umounted
455 	 * cleanly.  Finally, mark the quotas out of date so Windows rescans
456 	 * the volume on boot and updates them.
457 	 *
458 	 * When remounting read-only, mark the volume clean if no volume errors
459 	 * have occured.
460 	 */
461 	if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
462 		static const char *es = ".  Cannot remount read-write.";
463 
464 		/* Remounting read-write. */
465 		if (NVolErrors(vol)) {
466 			ntfs_error(sb, "Volume has errors and is read-only%s",
467 					es);
468 			return -EROFS;
469 		}
470 		if (vol->vol_flags & VOLUME_IS_DIRTY) {
471 			ntfs_error(sb, "Volume is dirty and read-only%s", es);
472 			return -EROFS;
473 		}
474 		if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
475 			ntfs_error(sb, "Volume has unsupported flags set and "
476 					"is read-only%s", es);
477 			return -EROFS;
478 		}
479 		if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
480 			ntfs_error(sb, "Failed to set dirty bit in volume "
481 					"information flags%s", es);
482 			return -EROFS;
483 		}
484 #if 0
485 		// TODO: Enable this code once we start modifying anything that
486 		//	 is different between NTFS 1.2 and 3.x...
487 		/* Set NT4 compatibility flag on newer NTFS version volumes. */
488 		if ((vol->major_ver > 1)) {
489 			if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
490 				ntfs_error(sb, "Failed to set NT4 "
491 						"compatibility flag%s", es);
492 				NVolSetErrors(vol);
493 				return -EROFS;
494 			}
495 		}
496 #endif
497 		if (!ntfs_empty_logfile(vol->logfile_ino)) {
498 			ntfs_error(sb, "Failed to empty journal $LogFile%s",
499 					es);
500 			NVolSetErrors(vol);
501 			return -EROFS;
502 		}
503 		if (!ntfs_mark_quotas_out_of_date(vol)) {
504 			ntfs_error(sb, "Failed to mark quotas out of date%s",
505 					es);
506 			NVolSetErrors(vol);
507 			return -EROFS;
508 		}
509 		if (!ntfs_stamp_usnjrnl(vol)) {
510 			ntfs_error(sb, "Failed to stamp transation log "
511 					"($UsnJrnl)%s", es);
512 			NVolSetErrors(vol);
513 			return -EROFS;
514 		}
515 	} else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
516 		/* Remounting read-only. */
517 		if (!NVolErrors(vol)) {
518 			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
519 				ntfs_warning(sb, "Failed to clear dirty bit "
520 						"in volume information "
521 						"flags.  Run chkdsk.");
522 		}
523 	}
524 #endif /* NTFS_RW */
525 
526 	// TODO: Deal with *flags.
527 
528 	if (!parse_options(vol, opt))
529 		return -EINVAL;
530 	ntfs_debug("Done.");
531 	return 0;
532 }
533 
534 /**
535  * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
536  * @sb:		Super block of the device to which @b belongs.
537  * @b:		Boot sector of device @sb to check.
538  * @silent:	If TRUE, all output will be silenced.
539  *
540  * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
541  * sector. Returns TRUE if it is valid and FALSE if not.
542  *
543  * @sb is only needed for warning/error output, i.e. it can be NULL when silent
544  * is TRUE.
545  */
546 static BOOL is_boot_sector_ntfs(const struct super_block *sb,
547 		const NTFS_BOOT_SECTOR *b, const BOOL silent)
548 {
549 	/*
550 	 * Check that checksum == sum of u32 values from b to the checksum
551 	 * field.  If checksum is zero, no checking is done.  We will work when
552 	 * the checksum test fails, since some utilities update the boot sector
553 	 * ignoring the checksum which leaves the checksum out-of-date.  We
554 	 * report a warning if this is the case.
555 	 */
556 	if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
557 		le32 *u;
558 		u32 i;
559 
560 		for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
561 			i += le32_to_cpup(u);
562 		if (le32_to_cpu(b->checksum) != i)
563 			ntfs_warning(sb, "Invalid boot sector checksum.");
564 	}
565 	/* Check OEMidentifier is "NTFS    " */
566 	if (b->oem_id != magicNTFS)
567 		goto not_ntfs;
568 	/* Check bytes per sector value is between 256 and 4096. */
569 	if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
570 			le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
571 		goto not_ntfs;
572 	/* Check sectors per cluster value is valid. */
573 	switch (b->bpb.sectors_per_cluster) {
574 	case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
575 		break;
576 	default:
577 		goto not_ntfs;
578 	}
579 	/* Check the cluster size is not above the maximum (64kiB). */
580 	if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
581 			b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
582 		goto not_ntfs;
583 	/* Check reserved/unused fields are really zero. */
584 	if (le16_to_cpu(b->bpb.reserved_sectors) ||
585 			le16_to_cpu(b->bpb.root_entries) ||
586 			le16_to_cpu(b->bpb.sectors) ||
587 			le16_to_cpu(b->bpb.sectors_per_fat) ||
588 			le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
589 		goto not_ntfs;
590 	/* Check clusters per file mft record value is valid. */
591 	if ((u8)b->clusters_per_mft_record < 0xe1 ||
592 			(u8)b->clusters_per_mft_record > 0xf7)
593 		switch (b->clusters_per_mft_record) {
594 		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
595 			break;
596 		default:
597 			goto not_ntfs;
598 		}
599 	/* Check clusters per index block value is valid. */
600 	if ((u8)b->clusters_per_index_record < 0xe1 ||
601 			(u8)b->clusters_per_index_record > 0xf7)
602 		switch (b->clusters_per_index_record) {
603 		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
604 			break;
605 		default:
606 			goto not_ntfs;
607 		}
608 	/*
609 	 * Check for valid end of sector marker. We will work without it, but
610 	 * many BIOSes will refuse to boot from a bootsector if the magic is
611 	 * incorrect, so we emit a warning.
612 	 */
613 	if (!silent && b->end_of_sector_marker != const_cpu_to_le16(0xaa55))
614 		ntfs_warning(sb, "Invalid end of sector marker.");
615 	return TRUE;
616 not_ntfs:
617 	return FALSE;
618 }
619 
620 /**
621  * read_ntfs_boot_sector - read the NTFS boot sector of a device
622  * @sb:		super block of device to read the boot sector from
623  * @silent:	if true, suppress all output
624  *
625  * Reads the boot sector from the device and validates it. If that fails, tries
626  * to read the backup boot sector, first from the end of the device a-la NT4 and
627  * later and then from the middle of the device a-la NT3.51 and before.
628  *
629  * If a valid boot sector is found but it is not the primary boot sector, we
630  * repair the primary boot sector silently (unless the device is read-only or
631  * the primary boot sector is not accessible).
632  *
633  * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
634  * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
635  * to their respective values.
636  *
637  * Return the unlocked buffer head containing the boot sector or NULL on error.
638  */
639 static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
640 		const int silent)
641 {
642 	const char *read_err_str = "Unable to read %s boot sector.";
643 	struct buffer_head *bh_primary, *bh_backup;
644 	long nr_blocks = NTFS_SB(sb)->nr_blocks;
645 
646 	/* Try to read primary boot sector. */
647 	if ((bh_primary = sb_bread(sb, 0))) {
648 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
649 				bh_primary->b_data, silent))
650 			return bh_primary;
651 		if (!silent)
652 			ntfs_error(sb, "Primary boot sector is invalid.");
653 	} else if (!silent)
654 		ntfs_error(sb, read_err_str, "primary");
655 	if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
656 		if (bh_primary)
657 			brelse(bh_primary);
658 		if (!silent)
659 			ntfs_error(sb, "Mount option errors=recover not used. "
660 					"Aborting without trying to recover.");
661 		return NULL;
662 	}
663 	/* Try to read NT4+ backup boot sector. */
664 	if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
665 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
666 				bh_backup->b_data, silent))
667 			goto hotfix_primary_boot_sector;
668 		brelse(bh_backup);
669 	} else if (!silent)
670 		ntfs_error(sb, read_err_str, "backup");
671 	/* Try to read NT3.51- backup boot sector. */
672 	if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
673 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
674 				bh_backup->b_data, silent))
675 			goto hotfix_primary_boot_sector;
676 		if (!silent)
677 			ntfs_error(sb, "Could not find a valid backup boot "
678 					"sector.");
679 		brelse(bh_backup);
680 	} else if (!silent)
681 		ntfs_error(sb, read_err_str, "backup");
682 	/* We failed. Cleanup and return. */
683 	if (bh_primary)
684 		brelse(bh_primary);
685 	return NULL;
686 hotfix_primary_boot_sector:
687 	if (bh_primary) {
688 		/*
689 		 * If we managed to read sector zero and the volume is not
690 		 * read-only, copy the found, valid backup boot sector to the
691 		 * primary boot sector.
692 		 */
693 		if (!(sb->s_flags & MS_RDONLY)) {
694 			ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
695 					"boot sector from backup copy.");
696 			memcpy(bh_primary->b_data, bh_backup->b_data,
697 					sb->s_blocksize);
698 			mark_buffer_dirty(bh_primary);
699 			sync_dirty_buffer(bh_primary);
700 			if (buffer_uptodate(bh_primary)) {
701 				brelse(bh_backup);
702 				return bh_primary;
703 			}
704 			ntfs_error(sb, "Hot-fix: Device write error while "
705 					"recovering primary boot sector.");
706 		} else {
707 			ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
708 					"sector failed: Read-only mount.");
709 		}
710 		brelse(bh_primary);
711 	}
712 	ntfs_warning(sb, "Using backup boot sector.");
713 	return bh_backup;
714 }
715 
716 /**
717  * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
718  * @vol:	volume structure to initialise with data from boot sector
719  * @b:		boot sector to parse
720  *
721  * Parse the ntfs boot sector @b and store all imporant information therein in
722  * the ntfs super block @vol.  Return TRUE on success and FALSE on error.
723  */
724 static BOOL parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
725 {
726 	unsigned int sectors_per_cluster_bits, nr_hidden_sects;
727 	int clusters_per_mft_record, clusters_per_index_record;
728 	s64 ll;
729 
730 	vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
731 	vol->sector_size_bits = ffs(vol->sector_size) - 1;
732 	ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
733 			vol->sector_size);
734 	ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
735 			vol->sector_size_bits);
736 	if (vol->sector_size != vol->sb->s_blocksize)
737 		ntfs_warning(vol->sb, "The boot sector indicates a sector size "
738 				"different from the device sector size.");
739 	ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
740 	sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
741 	ntfs_debug("sectors_per_cluster_bits = 0x%x",
742 			sectors_per_cluster_bits);
743 	nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
744 	ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
745 	vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
746 	vol->cluster_size_mask = vol->cluster_size - 1;
747 	vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
748 	ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
749 			vol->cluster_size);
750 	ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
751 	ntfs_debug("vol->cluster_size_bits = %i (0x%x)",
752 			vol->cluster_size_bits, vol->cluster_size_bits);
753 	if (vol->sector_size > vol->cluster_size) {
754 		ntfs_error(vol->sb, "Sector sizes above the cluster size are "
755 				"not supported.  Sorry.");
756 		return FALSE;
757 	}
758 	if (vol->sb->s_blocksize > vol->cluster_size) {
759 		ntfs_error(vol->sb, "Cluster sizes smaller than the device "
760 				"sector size are not supported.  Sorry.");
761 		return FALSE;
762 	}
763 	clusters_per_mft_record = b->clusters_per_mft_record;
764 	ntfs_debug("clusters_per_mft_record = %i (0x%x)",
765 			clusters_per_mft_record, clusters_per_mft_record);
766 	if (clusters_per_mft_record > 0)
767 		vol->mft_record_size = vol->cluster_size <<
768 				(ffs(clusters_per_mft_record) - 1);
769 	else
770 		/*
771 		 * When mft_record_size < cluster_size, clusters_per_mft_record
772 		 * = -log2(mft_record_size) bytes. mft_record_size normaly is
773 		 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
774 		 */
775 		vol->mft_record_size = 1 << -clusters_per_mft_record;
776 	vol->mft_record_size_mask = vol->mft_record_size - 1;
777 	vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
778 	ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
779 			vol->mft_record_size);
780 	ntfs_debug("vol->mft_record_size_mask = 0x%x",
781 			vol->mft_record_size_mask);
782 	ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
783 			vol->mft_record_size_bits, vol->mft_record_size_bits);
784 	/*
785 	 * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
786 	 * we store $MFT/$DATA, the table of mft records in the page cache.
787 	 */
788 	if (vol->mft_record_size > PAGE_CACHE_SIZE) {
789 		ntfs_error(vol->sb, "Mft record size %i (0x%x) exceeds the "
790 				"page cache size on your system %lu (0x%lx).  "
791 				"This is not supported.  Sorry.",
792 				vol->mft_record_size, vol->mft_record_size,
793 				PAGE_CACHE_SIZE, PAGE_CACHE_SIZE);
794 		return FALSE;
795 	}
796 	clusters_per_index_record = b->clusters_per_index_record;
797 	ntfs_debug("clusters_per_index_record = %i (0x%x)",
798 			clusters_per_index_record, clusters_per_index_record);
799 	if (clusters_per_index_record > 0)
800 		vol->index_record_size = vol->cluster_size <<
801 				(ffs(clusters_per_index_record) - 1);
802 	else
803 		/*
804 		 * When index_record_size < cluster_size,
805 		 * clusters_per_index_record = -log2(index_record_size) bytes.
806 		 * index_record_size normaly equals 4096 bytes, which is
807 		 * encoded as 0xF4 (-12 in decimal).
808 		 */
809 		vol->index_record_size = 1 << -clusters_per_index_record;
810 	vol->index_record_size_mask = vol->index_record_size - 1;
811 	vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
812 	ntfs_debug("vol->index_record_size = %i (0x%x)",
813 			vol->index_record_size, vol->index_record_size);
814 	ntfs_debug("vol->index_record_size_mask = 0x%x",
815 			vol->index_record_size_mask);
816 	ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
817 			vol->index_record_size_bits,
818 			vol->index_record_size_bits);
819 	/*
820 	 * Get the size of the volume in clusters and check for 64-bit-ness.
821 	 * Windows currently only uses 32 bits to save the clusters so we do
822 	 * the same as it is much faster on 32-bit CPUs.
823 	 */
824 	ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
825 	if ((u64)ll >= 1ULL << 32) {
826 		ntfs_error(vol->sb, "Cannot handle 64-bit clusters.  Sorry.");
827 		return FALSE;
828 	}
829 	vol->nr_clusters = ll;
830 	ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
831 	/*
832 	 * On an architecture where unsigned long is 32-bits, we restrict the
833 	 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
834 	 * will hopefully optimize the whole check away.
835 	 */
836 	if (sizeof(unsigned long) < 8) {
837 		if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
838 			ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
839 					"large for this architecture.  "
840 					"Maximum supported is 2TiB.  Sorry.",
841 					(unsigned long long)ll >> (40 -
842 					vol->cluster_size_bits));
843 			return FALSE;
844 		}
845 	}
846 	ll = sle64_to_cpu(b->mft_lcn);
847 	if (ll >= vol->nr_clusters) {
848 		ntfs_error(vol->sb, "MFT LCN is beyond end of volume.  Weird.");
849 		return FALSE;
850 	}
851 	vol->mft_lcn = ll;
852 	ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
853 	ll = sle64_to_cpu(b->mftmirr_lcn);
854 	if (ll >= vol->nr_clusters) {
855 		ntfs_error(vol->sb, "MFTMirr LCN is beyond end of volume.  "
856 				"Weird.");
857 		return FALSE;
858 	}
859 	vol->mftmirr_lcn = ll;
860 	ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
861 #ifdef NTFS_RW
862 	/*
863 	 * Work out the size of the mft mirror in number of mft records. If the
864 	 * cluster size is less than or equal to the size taken by four mft
865 	 * records, the mft mirror stores the first four mft records. If the
866 	 * cluster size is bigger than the size taken by four mft records, the
867 	 * mft mirror contains as many mft records as will fit into one
868 	 * cluster.
869 	 */
870 	if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
871 		vol->mftmirr_size = 4;
872 	else
873 		vol->mftmirr_size = vol->cluster_size >>
874 				vol->mft_record_size_bits;
875 	ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
876 #endif /* NTFS_RW */
877 	vol->serial_no = le64_to_cpu(b->volume_serial_number);
878 	ntfs_debug("vol->serial_no = 0x%llx",
879 			(unsigned long long)vol->serial_no);
880 	return TRUE;
881 }
882 
883 /**
884  * ntfs_setup_allocators - initialize the cluster and mft allocators
885  * @vol:	volume structure for which to setup the allocators
886  *
887  * Setup the cluster (lcn) and mft allocators to the starting values.
888  */
889 static void ntfs_setup_allocators(ntfs_volume *vol)
890 {
891 #ifdef NTFS_RW
892 	LCN mft_zone_size, mft_lcn;
893 #endif /* NTFS_RW */
894 
895 	ntfs_debug("vol->mft_zone_multiplier = 0x%x",
896 			vol->mft_zone_multiplier);
897 #ifdef NTFS_RW
898 	/* Determine the size of the MFT zone. */
899 	mft_zone_size = vol->nr_clusters;
900 	switch (vol->mft_zone_multiplier) {  /* % of volume size in clusters */
901 	case 4:
902 		mft_zone_size >>= 1;			/* 50%   */
903 		break;
904 	case 3:
905 		mft_zone_size = (mft_zone_size +
906 				(mft_zone_size >> 1)) >> 2;	/* 37.5% */
907 		break;
908 	case 2:
909 		mft_zone_size >>= 2;			/* 25%   */
910 		break;
911 	/* case 1: */
912 	default:
913 		mft_zone_size >>= 3;			/* 12.5% */
914 		break;
915 	}
916 	/* Setup the mft zone. */
917 	vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
918 	ntfs_debug("vol->mft_zone_pos = 0x%llx",
919 			(unsigned long long)vol->mft_zone_pos);
920 	/*
921 	 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
922 	 * source) and if the actual mft_lcn is in the expected place or even
923 	 * further to the front of the volume, extend the mft_zone to cover the
924 	 * beginning of the volume as well.  This is in order to protect the
925 	 * area reserved for the mft bitmap as well within the mft_zone itself.
926 	 * On non-standard volumes we do not protect it as the overhead would
927 	 * be higher than the speed increase we would get by doing it.
928 	 */
929 	mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
930 	if (mft_lcn * vol->cluster_size < 16 * 1024)
931 		mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
932 				vol->cluster_size;
933 	if (vol->mft_zone_start <= mft_lcn)
934 		vol->mft_zone_start = 0;
935 	ntfs_debug("vol->mft_zone_start = 0x%llx",
936 			(unsigned long long)vol->mft_zone_start);
937 	/*
938 	 * Need to cap the mft zone on non-standard volumes so that it does
939 	 * not point outside the boundaries of the volume.  We do this by
940 	 * halving the zone size until we are inside the volume.
941 	 */
942 	vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
943 	while (vol->mft_zone_end >= vol->nr_clusters) {
944 		mft_zone_size >>= 1;
945 		vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
946 	}
947 	ntfs_debug("vol->mft_zone_end = 0x%llx",
948 			(unsigned long long)vol->mft_zone_end);
949 	/*
950 	 * Set the current position within each data zone to the start of the
951 	 * respective zone.
952 	 */
953 	vol->data1_zone_pos = vol->mft_zone_end;
954 	ntfs_debug("vol->data1_zone_pos = 0x%llx",
955 			(unsigned long long)vol->data1_zone_pos);
956 	vol->data2_zone_pos = 0;
957 	ntfs_debug("vol->data2_zone_pos = 0x%llx",
958 			(unsigned long long)vol->data2_zone_pos);
959 
960 	/* Set the mft data allocation position to mft record 24. */
961 	vol->mft_data_pos = 24;
962 	ntfs_debug("vol->mft_data_pos = 0x%llx",
963 			(unsigned long long)vol->mft_data_pos);
964 #endif /* NTFS_RW */
965 }
966 
967 #ifdef NTFS_RW
968 
969 /**
970  * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
971  * @vol:	ntfs super block describing device whose mft mirror to load
972  *
973  * Return TRUE on success or FALSE on error.
974  */
975 static BOOL load_and_init_mft_mirror(ntfs_volume *vol)
976 {
977 	struct inode *tmp_ino;
978 	ntfs_inode *tmp_ni;
979 
980 	ntfs_debug("Entering.");
981 	/* Get mft mirror inode. */
982 	tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
983 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
984 		if (!IS_ERR(tmp_ino))
985 			iput(tmp_ino);
986 		/* Caller will display error message. */
987 		return FALSE;
988 	}
989 	/*
990 	 * Re-initialize some specifics about $MFTMirr's inode as
991 	 * ntfs_read_inode() will have set up the default ones.
992 	 */
993 	/* Set uid and gid to root. */
994 	tmp_ino->i_uid = tmp_ino->i_gid = 0;
995 	/* Regular file.  No access for anyone. */
996 	tmp_ino->i_mode = S_IFREG;
997 	/* No VFS initiated operations allowed for $MFTMirr. */
998 	tmp_ino->i_op = &ntfs_empty_inode_ops;
999 	tmp_ino->i_fop = &ntfs_empty_file_ops;
1000 	/* Put in our special address space operations. */
1001 	tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1002 	tmp_ni = NTFS_I(tmp_ino);
1003 	/* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1004 	NInoSetMstProtected(tmp_ni);
1005 	NInoSetSparseDisabled(tmp_ni);
1006 	/*
1007 	 * Set up our little cheat allowing us to reuse the async read io
1008 	 * completion handler for directories.
1009 	 */
1010 	tmp_ni->itype.index.block_size = vol->mft_record_size;
1011 	tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1012 	vol->mftmirr_ino = tmp_ino;
1013 	ntfs_debug("Done.");
1014 	return TRUE;
1015 }
1016 
1017 /**
1018  * check_mft_mirror - compare contents of the mft mirror with the mft
1019  * @vol:	ntfs super block describing device whose mft mirror to check
1020  *
1021  * Return TRUE on success or FALSE on error.
1022  *
1023  * Note, this function also results in the mft mirror runlist being completely
1024  * mapped into memory.  The mft mirror write code requires this and will BUG()
1025  * should it find an unmapped runlist element.
1026  */
1027 static BOOL check_mft_mirror(ntfs_volume *vol)
1028 {
1029 	struct super_block *sb = vol->sb;
1030 	ntfs_inode *mirr_ni;
1031 	struct page *mft_page, *mirr_page;
1032 	u8 *kmft, *kmirr;
1033 	runlist_element *rl, rl2[2];
1034 	pgoff_t index;
1035 	int mrecs_per_page, i;
1036 
1037 	ntfs_debug("Entering.");
1038 	/* Compare contents of $MFT and $MFTMirr. */
1039 	mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1040 	BUG_ON(!mrecs_per_page);
1041 	BUG_ON(!vol->mftmirr_size);
1042 	mft_page = mirr_page = NULL;
1043 	kmft = kmirr = NULL;
1044 	index = i = 0;
1045 	do {
1046 		u32 bytes;
1047 
1048 		/* Switch pages if necessary. */
1049 		if (!(i % mrecs_per_page)) {
1050 			if (index) {
1051 				ntfs_unmap_page(mft_page);
1052 				ntfs_unmap_page(mirr_page);
1053 			}
1054 			/* Get the $MFT page. */
1055 			mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1056 					index);
1057 			if (IS_ERR(mft_page)) {
1058 				ntfs_error(sb, "Failed to read $MFT.");
1059 				return FALSE;
1060 			}
1061 			kmft = page_address(mft_page);
1062 			/* Get the $MFTMirr page. */
1063 			mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1064 					index);
1065 			if (IS_ERR(mirr_page)) {
1066 				ntfs_error(sb, "Failed to read $MFTMirr.");
1067 				goto mft_unmap_out;
1068 			}
1069 			kmirr = page_address(mirr_page);
1070 			++index;
1071 		}
1072 		/* Make sure the record is ok. */
1073 		if (ntfs_is_baad_recordp((le32*)kmft)) {
1074 			ntfs_error(sb, "Incomplete multi sector transfer "
1075 					"detected in mft record %i.", i);
1076 mm_unmap_out:
1077 			ntfs_unmap_page(mirr_page);
1078 mft_unmap_out:
1079 			ntfs_unmap_page(mft_page);
1080 			return FALSE;
1081 		}
1082 		if (ntfs_is_baad_recordp((le32*)kmirr)) {
1083 			ntfs_error(sb, "Incomplete multi sector transfer "
1084 					"detected in mft mirror record %i.", i);
1085 			goto mm_unmap_out;
1086 		}
1087 		/* Get the amount of data in the current record. */
1088 		bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1089 		if (!bytes || bytes > vol->mft_record_size) {
1090 			bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1091 			if (!bytes || bytes > vol->mft_record_size)
1092 				bytes = vol->mft_record_size;
1093 		}
1094 		/* Compare the two records. */
1095 		if (memcmp(kmft, kmirr, bytes)) {
1096 			ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1097 					"match.  Run ntfsfix or chkdsk.", i);
1098 			goto mm_unmap_out;
1099 		}
1100 		kmft += vol->mft_record_size;
1101 		kmirr += vol->mft_record_size;
1102 	} while (++i < vol->mftmirr_size);
1103 	/* Release the last pages. */
1104 	ntfs_unmap_page(mft_page);
1105 	ntfs_unmap_page(mirr_page);
1106 
1107 	/* Construct the mft mirror runlist by hand. */
1108 	rl2[0].vcn = 0;
1109 	rl2[0].lcn = vol->mftmirr_lcn;
1110 	rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1111 			vol->cluster_size - 1) / vol->cluster_size;
1112 	rl2[1].vcn = rl2[0].length;
1113 	rl2[1].lcn = LCN_ENOENT;
1114 	rl2[1].length = 0;
1115 	/*
1116 	 * Because we have just read all of the mft mirror, we know we have
1117 	 * mapped the full runlist for it.
1118 	 */
1119 	mirr_ni = NTFS_I(vol->mftmirr_ino);
1120 	down_read(&mirr_ni->runlist.lock);
1121 	rl = mirr_ni->runlist.rl;
1122 	/* Compare the two runlists.  They must be identical. */
1123 	i = 0;
1124 	do {
1125 		if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1126 				rl2[i].length != rl[i].length) {
1127 			ntfs_error(sb, "$MFTMirr location mismatch.  "
1128 					"Run chkdsk.");
1129 			up_read(&mirr_ni->runlist.lock);
1130 			return FALSE;
1131 		}
1132 	} while (rl2[i++].length);
1133 	up_read(&mirr_ni->runlist.lock);
1134 	ntfs_debug("Done.");
1135 	return TRUE;
1136 }
1137 
1138 /**
1139  * load_and_check_logfile - load and check the logfile inode for a volume
1140  * @vol:	ntfs super block describing device whose logfile to load
1141  *
1142  * Return TRUE on success or FALSE on error.
1143  */
1144 static BOOL load_and_check_logfile(ntfs_volume *vol,
1145 		RESTART_PAGE_HEADER **rp)
1146 {
1147 	struct inode *tmp_ino;
1148 
1149 	ntfs_debug("Entering.");
1150 	tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1151 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1152 		if (!IS_ERR(tmp_ino))
1153 			iput(tmp_ino);
1154 		/* Caller will display error message. */
1155 		return FALSE;
1156 	}
1157 	if (!ntfs_check_logfile(tmp_ino, rp)) {
1158 		iput(tmp_ino);
1159 		/* ntfs_check_logfile() will have displayed error output. */
1160 		return FALSE;
1161 	}
1162 	NInoSetSparseDisabled(NTFS_I(tmp_ino));
1163 	vol->logfile_ino = tmp_ino;
1164 	ntfs_debug("Done.");
1165 	return TRUE;
1166 }
1167 
1168 #define NTFS_HIBERFIL_HEADER_SIZE	4096
1169 
1170 /**
1171  * check_windows_hibernation_status - check if Windows is suspended on a volume
1172  * @vol:	ntfs super block of device to check
1173  *
1174  * Check if Windows is hibernated on the ntfs volume @vol.  This is done by
1175  * looking for the file hiberfil.sys in the root directory of the volume.  If
1176  * the file is not present Windows is definitely not suspended.
1177  *
1178  * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1179  * definitely suspended (this volume is not the system volume).  Caveat:  on a
1180  * system with many volumes it is possible that the < 4kiB check is bogus but
1181  * for now this should do fine.
1182  *
1183  * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1184  * hiberfil header (which is the first 4kiB).  If this begins with "hibr",
1185  * Windows is definitely suspended.  If it is completely full of zeroes,
1186  * Windows is definitely not hibernated.  Any other case is treated as if
1187  * Windows is suspended.  This caters for the above mentioned caveat of a
1188  * system with many volumes where no "hibr" magic would be present and there is
1189  * no zero header.
1190  *
1191  * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1192  * hibernated on the volume, and -errno on error.
1193  */
1194 static int check_windows_hibernation_status(ntfs_volume *vol)
1195 {
1196 	MFT_REF mref;
1197 	struct inode *vi;
1198 	ntfs_inode *ni;
1199 	struct page *page;
1200 	u32 *kaddr, *kend;
1201 	ntfs_name *name = NULL;
1202 	int ret = 1;
1203 	static const ntfschar hiberfil[13] = { const_cpu_to_le16('h'),
1204 			const_cpu_to_le16('i'), const_cpu_to_le16('b'),
1205 			const_cpu_to_le16('e'), const_cpu_to_le16('r'),
1206 			const_cpu_to_le16('f'), const_cpu_to_le16('i'),
1207 			const_cpu_to_le16('l'), const_cpu_to_le16('.'),
1208 			const_cpu_to_le16('s'), const_cpu_to_le16('y'),
1209 			const_cpu_to_le16('s'), 0 };
1210 
1211 	ntfs_debug("Entering.");
1212 	/*
1213 	 * Find the inode number for the hibernation file by looking up the
1214 	 * filename hiberfil.sys in the root directory.
1215 	 */
1216 	down(&vol->root_ino->i_sem);
1217 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1218 			&name);
1219 	up(&vol->root_ino->i_sem);
1220 	if (IS_ERR_MREF(mref)) {
1221 		ret = MREF_ERR(mref);
1222 		/* If the file does not exist, Windows is not hibernated. */
1223 		if (ret == -ENOENT) {
1224 			ntfs_debug("hiberfil.sys not present.  Windows is not "
1225 					"hibernated on the volume.");
1226 			return 0;
1227 		}
1228 		/* A real error occured. */
1229 		ntfs_error(vol->sb, "Failed to find inode number for "
1230 				"hiberfil.sys.");
1231 		return ret;
1232 	}
1233 	/* We do not care for the type of match that was found. */
1234 	kfree(name);
1235 	/* Get the inode. */
1236 	vi = ntfs_iget(vol->sb, MREF(mref));
1237 	if (IS_ERR(vi) || is_bad_inode(vi)) {
1238 		if (!IS_ERR(vi))
1239 			iput(vi);
1240 		ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1241 		return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1242 	}
1243 	if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1244 		ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx).  "
1245 				"Windows is hibernated on the volume.  This "
1246 				"is not the system volume.", i_size_read(vi));
1247 		goto iput_out;
1248 	}
1249 	ni = NTFS_I(vi);
1250 	page = ntfs_map_page(vi->i_mapping, 0);
1251 	if (IS_ERR(page)) {
1252 		ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1253 		ret = PTR_ERR(page);
1254 		goto iput_out;
1255 	}
1256 	kaddr = (u32*)page_address(page);
1257 	if (*(le32*)kaddr == const_cpu_to_le32(0x72626968)/*'hibr'*/) {
1258 		ntfs_debug("Magic \"hibr\" found in hiberfil.sys.  Windows is "
1259 				"hibernated on the volume.  This is the "
1260 				"system volume.");
1261 		goto unm_iput_out;
1262 	}
1263 	kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1264 	do {
1265 		if (unlikely(*kaddr)) {
1266 			ntfs_debug("hiberfil.sys is larger than 4kiB "
1267 					"(0x%llx), does not contain the "
1268 					"\"hibr\" magic, and does not have a "
1269 					"zero header.  Windows is hibernated "
1270 					"on the volume.  This is not the "
1271 					"system volume.", i_size_read(vi));
1272 			goto unm_iput_out;
1273 		}
1274 	} while (++kaddr < kend);
1275 	ntfs_debug("hiberfil.sys contains a zero header.  Windows is not "
1276 			"hibernated on the volume.  This is the system "
1277 			"volume.");
1278 	ret = 0;
1279 unm_iput_out:
1280 	ntfs_unmap_page(page);
1281 iput_out:
1282 	iput(vi);
1283 	return ret;
1284 }
1285 
1286 /**
1287  * load_and_init_quota - load and setup the quota file for a volume if present
1288  * @vol:	ntfs super block describing device whose quota file to load
1289  *
1290  * Return TRUE on success or FALSE on error.  If $Quota is not present, we
1291  * leave vol->quota_ino as NULL and return success.
1292  */
1293 static BOOL load_and_init_quota(ntfs_volume *vol)
1294 {
1295 	MFT_REF mref;
1296 	struct inode *tmp_ino;
1297 	ntfs_name *name = NULL;
1298 	static const ntfschar Quota[7] = { const_cpu_to_le16('$'),
1299 			const_cpu_to_le16('Q'), const_cpu_to_le16('u'),
1300 			const_cpu_to_le16('o'), const_cpu_to_le16('t'),
1301 			const_cpu_to_le16('a'), 0 };
1302 	static ntfschar Q[3] = { const_cpu_to_le16('$'),
1303 			const_cpu_to_le16('Q'), 0 };
1304 
1305 	ntfs_debug("Entering.");
1306 	/*
1307 	 * Find the inode number for the quota file by looking up the filename
1308 	 * $Quota in the extended system files directory $Extend.
1309 	 */
1310 	down(&vol->extend_ino->i_sem);
1311 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1312 			&name);
1313 	up(&vol->extend_ino->i_sem);
1314 	if (IS_ERR_MREF(mref)) {
1315 		/*
1316 		 * If the file does not exist, quotas are disabled and have
1317 		 * never been enabled on this volume, just return success.
1318 		 */
1319 		if (MREF_ERR(mref) == -ENOENT) {
1320 			ntfs_debug("$Quota not present.  Volume does not have "
1321 					"quotas enabled.");
1322 			/*
1323 			 * No need to try to set quotas out of date if they are
1324 			 * not enabled.
1325 			 */
1326 			NVolSetQuotaOutOfDate(vol);
1327 			return TRUE;
1328 		}
1329 		/* A real error occured. */
1330 		ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1331 		return FALSE;
1332 	}
1333 	/* We do not care for the type of match that was found. */
1334 	kfree(name);
1335 	/* Get the inode. */
1336 	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1337 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1338 		if (!IS_ERR(tmp_ino))
1339 			iput(tmp_ino);
1340 		ntfs_error(vol->sb, "Failed to load $Quota.");
1341 		return FALSE;
1342 	}
1343 	vol->quota_ino = tmp_ino;
1344 	/* Get the $Q index allocation attribute. */
1345 	tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1346 	if (IS_ERR(tmp_ino)) {
1347 		ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1348 		return FALSE;
1349 	}
1350 	vol->quota_q_ino = tmp_ino;
1351 	ntfs_debug("Done.");
1352 	return TRUE;
1353 }
1354 
1355 /**
1356  * load_and_init_usnjrnl - load and setup the transaction log if present
1357  * @vol:	ntfs super block describing device whose usnjrnl file to load
1358  *
1359  * Return TRUE on success or FALSE on error.
1360  *
1361  * If $UsnJrnl is not present or in the process of being disabled, we set
1362  * NVolUsnJrnlStamped() and return success.
1363  *
1364  * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1365  * i.e. transaction logging has only just been enabled or the journal has been
1366  * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1367  * and return success.
1368  */
1369 static BOOL load_and_init_usnjrnl(ntfs_volume *vol)
1370 {
1371 	MFT_REF mref;
1372 	struct inode *tmp_ino;
1373 	ntfs_inode *tmp_ni;
1374 	struct page *page;
1375 	ntfs_name *name = NULL;
1376 	USN_HEADER *uh;
1377 	static const ntfschar UsnJrnl[9] = { const_cpu_to_le16('$'),
1378 			const_cpu_to_le16('U'), const_cpu_to_le16('s'),
1379 			const_cpu_to_le16('n'), const_cpu_to_le16('J'),
1380 			const_cpu_to_le16('r'), const_cpu_to_le16('n'),
1381 			const_cpu_to_le16('l'), 0 };
1382 	static ntfschar Max[5] = { const_cpu_to_le16('$'),
1383 			const_cpu_to_le16('M'), const_cpu_to_le16('a'),
1384 			const_cpu_to_le16('x'), 0 };
1385 	static ntfschar J[3] = { const_cpu_to_le16('$'),
1386 			const_cpu_to_le16('J'), 0 };
1387 
1388 	ntfs_debug("Entering.");
1389 	/*
1390 	 * Find the inode number for the transaction log file by looking up the
1391 	 * filename $UsnJrnl in the extended system files directory $Extend.
1392 	 */
1393 	down(&vol->extend_ino->i_sem);
1394 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1395 			&name);
1396 	up(&vol->extend_ino->i_sem);
1397 	if (IS_ERR_MREF(mref)) {
1398 		/*
1399 		 * If the file does not exist, transaction logging is disabled,
1400 		 * just return success.
1401 		 */
1402 		if (MREF_ERR(mref) == -ENOENT) {
1403 			ntfs_debug("$UsnJrnl not present.  Volume does not "
1404 					"have transaction logging enabled.");
1405 not_enabled:
1406 			/*
1407 			 * No need to try to stamp the transaction log if
1408 			 * transaction logging is not enabled.
1409 			 */
1410 			NVolSetUsnJrnlStamped(vol);
1411 			return TRUE;
1412 		}
1413 		/* A real error occured. */
1414 		ntfs_error(vol->sb, "Failed to find inode number for "
1415 				"$UsnJrnl.");
1416 		return FALSE;
1417 	}
1418 	/* We do not care for the type of match that was found. */
1419 	kfree(name);
1420 	/* Get the inode. */
1421 	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1422 	if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1423 		if (!IS_ERR(tmp_ino))
1424 			iput(tmp_ino);
1425 		ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1426 		return FALSE;
1427 	}
1428 	vol->usnjrnl_ino = tmp_ino;
1429 	/*
1430 	 * If the transaction log is in the process of being deleted, we can
1431 	 * ignore it.
1432 	 */
1433 	if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1434 		ntfs_debug("$UsnJrnl in the process of being disabled.  "
1435 				"Volume does not have transaction logging "
1436 				"enabled.");
1437 		goto not_enabled;
1438 	}
1439 	/* Get the $DATA/$Max attribute. */
1440 	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1441 	if (IS_ERR(tmp_ino)) {
1442 		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1443 				"attribute.");
1444 		return FALSE;
1445 	}
1446 	vol->usnjrnl_max_ino = tmp_ino;
1447 	if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1448 		ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1449 				"attribute (size is 0x%llx but should be at "
1450 				"least 0x%x bytes).", i_size_read(tmp_ino),
1451 				sizeof(USN_HEADER));
1452 		return FALSE;
1453 	}
1454 	/* Get the $DATA/$J attribute. */
1455 	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1456 	if (IS_ERR(tmp_ino)) {
1457 		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1458 				"attribute.");
1459 		return FALSE;
1460 	}
1461 	vol->usnjrnl_j_ino = tmp_ino;
1462 	/* Verify $J is non-resident and sparse. */
1463 	tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1464 	if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1465 		ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1466 				"and/or not sparse.");
1467 		return FALSE;
1468 	}
1469 	/* Read the USN_HEADER from $DATA/$Max. */
1470 	page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1471 	if (IS_ERR(page)) {
1472 		ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1473 				"attribute.");
1474 		return FALSE;
1475 	}
1476 	uh = (USN_HEADER*)page_address(page);
1477 	/* Sanity check the $Max. */
1478 	if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1479 			sle64_to_cpu(uh->maximum_size))) {
1480 		ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1481 				"maximum size (0x%llx).  $UsnJrnl is corrupt.",
1482 				(long long)sle64_to_cpu(uh->allocation_delta),
1483 				(long long)sle64_to_cpu(uh->maximum_size));
1484 		ntfs_unmap_page(page);
1485 		return FALSE;
1486 	}
1487 	/*
1488 	 * If the transaction log has been stamped and nothing has been written
1489 	 * to it since, we do not need to stamp it.
1490 	 */
1491 	if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1492 			i_size_read(vol->usnjrnl_j_ino))) {
1493 		if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1494 				i_size_read(vol->usnjrnl_j_ino))) {
1495 			ntfs_unmap_page(page);
1496 			ntfs_debug("$UsnJrnl is enabled but nothing has been "
1497 					"logged since it was last stamped.  "
1498 					"Treating this as if the volume does "
1499 					"not have transaction logging "
1500 					"enabled.");
1501 			goto not_enabled;
1502 		}
1503 		ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1504 				"which is out of bounds (0x%llx).  $UsnJrnl "
1505 				"is corrupt.",
1506 				(long long)sle64_to_cpu(uh->lowest_valid_usn),
1507 				i_size_read(vol->usnjrnl_j_ino));
1508 		ntfs_unmap_page(page);
1509 		return FALSE;
1510 	}
1511 	ntfs_unmap_page(page);
1512 	ntfs_debug("Done.");
1513 	return TRUE;
1514 }
1515 
1516 /**
1517  * load_and_init_attrdef - load the attribute definitions table for a volume
1518  * @vol:	ntfs super block describing device whose attrdef to load
1519  *
1520  * Return TRUE on success or FALSE on error.
1521  */
1522 static BOOL load_and_init_attrdef(ntfs_volume *vol)
1523 {
1524 	loff_t i_size;
1525 	struct super_block *sb = vol->sb;
1526 	struct inode *ino;
1527 	struct page *page;
1528 	pgoff_t index, max_index;
1529 	unsigned int size;
1530 
1531 	ntfs_debug("Entering.");
1532 	/* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1533 	ino = ntfs_iget(sb, FILE_AttrDef);
1534 	if (IS_ERR(ino) || is_bad_inode(ino)) {
1535 		if (!IS_ERR(ino))
1536 			iput(ino);
1537 		goto failed;
1538 	}
1539 	NInoSetSparseDisabled(NTFS_I(ino));
1540 	/* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1541 	i_size = i_size_read(ino);
1542 	if (i_size <= 0 || i_size > 0x7fffffff)
1543 		goto iput_failed;
1544 	vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1545 	if (!vol->attrdef)
1546 		goto iput_failed;
1547 	index = 0;
1548 	max_index = i_size >> PAGE_CACHE_SHIFT;
1549 	size = PAGE_CACHE_SIZE;
1550 	while (index < max_index) {
1551 		/* Read the attrdef table and copy it into the linear buffer. */
1552 read_partial_attrdef_page:
1553 		page = ntfs_map_page(ino->i_mapping, index);
1554 		if (IS_ERR(page))
1555 			goto free_iput_failed;
1556 		memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1557 				page_address(page), size);
1558 		ntfs_unmap_page(page);
1559 	};
1560 	if (size == PAGE_CACHE_SIZE) {
1561 		size = i_size & ~PAGE_CACHE_MASK;
1562 		if (size)
1563 			goto read_partial_attrdef_page;
1564 	}
1565 	vol->attrdef_size = i_size;
1566 	ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1567 	iput(ino);
1568 	return TRUE;
1569 free_iput_failed:
1570 	ntfs_free(vol->attrdef);
1571 	vol->attrdef = NULL;
1572 iput_failed:
1573 	iput(ino);
1574 failed:
1575 	ntfs_error(sb, "Failed to initialize attribute definition table.");
1576 	return FALSE;
1577 }
1578 
1579 #endif /* NTFS_RW */
1580 
1581 /**
1582  * load_and_init_upcase - load the upcase table for an ntfs volume
1583  * @vol:	ntfs super block describing device whose upcase to load
1584  *
1585  * Return TRUE on success or FALSE on error.
1586  */
1587 static BOOL load_and_init_upcase(ntfs_volume *vol)
1588 {
1589 	loff_t i_size;
1590 	struct super_block *sb = vol->sb;
1591 	struct inode *ino;
1592 	struct page *page;
1593 	pgoff_t index, max_index;
1594 	unsigned int size;
1595 	int i, max;
1596 
1597 	ntfs_debug("Entering.");
1598 	/* Read upcase table and setup vol->upcase and vol->upcase_len. */
1599 	ino = ntfs_iget(sb, FILE_UpCase);
1600 	if (IS_ERR(ino) || is_bad_inode(ino)) {
1601 		if (!IS_ERR(ino))
1602 			iput(ino);
1603 		goto upcase_failed;
1604 	}
1605 	/*
1606 	 * The upcase size must not be above 64k Unicode characters, must not
1607 	 * be zero and must be a multiple of sizeof(ntfschar).
1608 	 */
1609 	i_size = i_size_read(ino);
1610 	if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1611 			i_size > 64ULL * 1024 * sizeof(ntfschar))
1612 		goto iput_upcase_failed;
1613 	vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1614 	if (!vol->upcase)
1615 		goto iput_upcase_failed;
1616 	index = 0;
1617 	max_index = i_size >> PAGE_CACHE_SHIFT;
1618 	size = PAGE_CACHE_SIZE;
1619 	while (index < max_index) {
1620 		/* Read the upcase table and copy it into the linear buffer. */
1621 read_partial_upcase_page:
1622 		page = ntfs_map_page(ino->i_mapping, index);
1623 		if (IS_ERR(page))
1624 			goto iput_upcase_failed;
1625 		memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1626 				page_address(page), size);
1627 		ntfs_unmap_page(page);
1628 	};
1629 	if (size == PAGE_CACHE_SIZE) {
1630 		size = i_size & ~PAGE_CACHE_MASK;
1631 		if (size)
1632 			goto read_partial_upcase_page;
1633 	}
1634 	vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1635 	ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1636 			i_size, 64 * 1024 * sizeof(ntfschar));
1637 	iput(ino);
1638 	down(&ntfs_lock);
1639 	if (!default_upcase) {
1640 		ntfs_debug("Using volume specified $UpCase since default is "
1641 				"not present.");
1642 		up(&ntfs_lock);
1643 		return TRUE;
1644 	}
1645 	max = default_upcase_len;
1646 	if (max > vol->upcase_len)
1647 		max = vol->upcase_len;
1648 	for (i = 0; i < max; i++)
1649 		if (vol->upcase[i] != default_upcase[i])
1650 			break;
1651 	if (i == max) {
1652 		ntfs_free(vol->upcase);
1653 		vol->upcase = default_upcase;
1654 		vol->upcase_len = max;
1655 		ntfs_nr_upcase_users++;
1656 		up(&ntfs_lock);
1657 		ntfs_debug("Volume specified $UpCase matches default. Using "
1658 				"default.");
1659 		return TRUE;
1660 	}
1661 	up(&ntfs_lock);
1662 	ntfs_debug("Using volume specified $UpCase since it does not match "
1663 			"the default.");
1664 	return TRUE;
1665 iput_upcase_failed:
1666 	iput(ino);
1667 	ntfs_free(vol->upcase);
1668 	vol->upcase = NULL;
1669 upcase_failed:
1670 	down(&ntfs_lock);
1671 	if (default_upcase) {
1672 		vol->upcase = default_upcase;
1673 		vol->upcase_len = default_upcase_len;
1674 		ntfs_nr_upcase_users++;
1675 		up(&ntfs_lock);
1676 		ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1677 				"default.");
1678 		return TRUE;
1679 	}
1680 	up(&ntfs_lock);
1681 	ntfs_error(sb, "Failed to initialize upcase table.");
1682 	return FALSE;
1683 }
1684 
1685 /**
1686  * load_system_files - open the system files using normal functions
1687  * @vol:	ntfs super block describing device whose system files to load
1688  *
1689  * Open the system files with normal access functions and complete setting up
1690  * the ntfs super block @vol.
1691  *
1692  * Return TRUE on success or FALSE on error.
1693  */
1694 static BOOL load_system_files(ntfs_volume *vol)
1695 {
1696 	struct super_block *sb = vol->sb;
1697 	MFT_RECORD *m;
1698 	VOLUME_INFORMATION *vi;
1699 	ntfs_attr_search_ctx *ctx;
1700 #ifdef NTFS_RW
1701 	RESTART_PAGE_HEADER *rp;
1702 	int err;
1703 #endif /* NTFS_RW */
1704 
1705 	ntfs_debug("Entering.");
1706 #ifdef NTFS_RW
1707 	/* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1708 	if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1709 		static const char *es1 = "Failed to load $MFTMirr";
1710 		static const char *es2 = "$MFTMirr does not match $MFT";
1711 		static const char *es3 = ".  Run ntfsfix and/or chkdsk.";
1712 
1713 		/* If a read-write mount, convert it to a read-only mount. */
1714 		if (!(sb->s_flags & MS_RDONLY)) {
1715 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1716 					ON_ERRORS_CONTINUE))) {
1717 				ntfs_error(sb, "%s and neither on_errors="
1718 						"continue nor on_errors="
1719 						"remount-ro was specified%s",
1720 						!vol->mftmirr_ino ? es1 : es2,
1721 						es3);
1722 				goto iput_mirr_err_out;
1723 			}
1724 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1725 			ntfs_error(sb, "%s.  Mounting read-only%s",
1726 					!vol->mftmirr_ino ? es1 : es2, es3);
1727 		} else
1728 			ntfs_warning(sb, "%s.  Will not be able to remount "
1729 					"read-write%s",
1730 					!vol->mftmirr_ino ? es1 : es2, es3);
1731 		/* This will prevent a read-write remount. */
1732 		NVolSetErrors(vol);
1733 	}
1734 #endif /* NTFS_RW */
1735 	/* Get mft bitmap attribute inode. */
1736 	vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1737 	if (IS_ERR(vol->mftbmp_ino)) {
1738 		ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1739 		goto iput_mirr_err_out;
1740 	}
1741 	/* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1742 	if (!load_and_init_upcase(vol))
1743 		goto iput_mftbmp_err_out;
1744 #ifdef NTFS_RW
1745 	/*
1746 	 * Read attribute definitions table and setup @vol->attrdef and
1747 	 * @vol->attrdef_size.
1748 	 */
1749 	if (!load_and_init_attrdef(vol))
1750 		goto iput_upcase_err_out;
1751 #endif /* NTFS_RW */
1752 	/*
1753 	 * Get the cluster allocation bitmap inode and verify the size, no
1754 	 * need for any locking at this stage as we are already running
1755 	 * exclusively as we are mount in progress task.
1756 	 */
1757 	vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1758 	if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1759 		if (!IS_ERR(vol->lcnbmp_ino))
1760 			iput(vol->lcnbmp_ino);
1761 		goto bitmap_failed;
1762 	}
1763 	NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1764 	if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1765 		iput(vol->lcnbmp_ino);
1766 bitmap_failed:
1767 		ntfs_error(sb, "Failed to load $Bitmap.");
1768 		goto iput_attrdef_err_out;
1769 	}
1770 	/*
1771 	 * Get the volume inode and setup our cache of the volume flags and
1772 	 * version.
1773 	 */
1774 	vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1775 	if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1776 		if (!IS_ERR(vol->vol_ino))
1777 			iput(vol->vol_ino);
1778 volume_failed:
1779 		ntfs_error(sb, "Failed to load $Volume.");
1780 		goto iput_lcnbmp_err_out;
1781 	}
1782 	m = map_mft_record(NTFS_I(vol->vol_ino));
1783 	if (IS_ERR(m)) {
1784 iput_volume_failed:
1785 		iput(vol->vol_ino);
1786 		goto volume_failed;
1787 	}
1788 	if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1789 		ntfs_error(sb, "Failed to get attribute search context.");
1790 		goto get_ctx_vol_failed;
1791 	}
1792 	if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1793 			ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1794 err_put_vol:
1795 		ntfs_attr_put_search_ctx(ctx);
1796 get_ctx_vol_failed:
1797 		unmap_mft_record(NTFS_I(vol->vol_ino));
1798 		goto iput_volume_failed;
1799 	}
1800 	vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1801 			le16_to_cpu(ctx->attr->data.resident.value_offset));
1802 	/* Some bounds checks. */
1803 	if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1804 			le32_to_cpu(ctx->attr->data.resident.value_length) >
1805 			(u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1806 		goto err_put_vol;
1807 	/* Copy the volume flags and version to the ntfs_volume structure. */
1808 	vol->vol_flags = vi->flags;
1809 	vol->major_ver = vi->major_ver;
1810 	vol->minor_ver = vi->minor_ver;
1811 	ntfs_attr_put_search_ctx(ctx);
1812 	unmap_mft_record(NTFS_I(vol->vol_ino));
1813 	printk(KERN_INFO "NTFS volume version %i.%i.\n", vol->major_ver,
1814 			vol->minor_ver);
1815 	if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1816 		ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1817 				"volume version %i.%i (need at least version "
1818 				"3.0).", vol->major_ver, vol->minor_ver);
1819 		NVolClearSparseEnabled(vol);
1820 	}
1821 #ifdef NTFS_RW
1822 	/* Make sure that no unsupported volume flags are set. */
1823 	if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1824 		static const char *es1a = "Volume is dirty";
1825 		static const char *es1b = "Volume has unsupported flags set";
1826 		static const char *es2 = ".  Run chkdsk and mount in Windows.";
1827 		const char *es1;
1828 
1829 		es1 = vol->vol_flags & VOLUME_IS_DIRTY ? es1a : es1b;
1830 		/* If a read-write mount, convert it to a read-only mount. */
1831 		if (!(sb->s_flags & MS_RDONLY)) {
1832 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1833 					ON_ERRORS_CONTINUE))) {
1834 				ntfs_error(sb, "%s and neither on_errors="
1835 						"continue nor on_errors="
1836 						"remount-ro was specified%s",
1837 						es1, es2);
1838 				goto iput_vol_err_out;
1839 			}
1840 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1841 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1842 		} else
1843 			ntfs_warning(sb, "%s.  Will not be able to remount "
1844 					"read-write%s", es1, es2);
1845 		/*
1846 		 * Do not set NVolErrors() because ntfs_remount() re-checks the
1847 		 * flags which we need to do in case any flags have changed.
1848 		 */
1849 	}
1850 	/*
1851 	 * Get the inode for the logfile, check it and determine if the volume
1852 	 * was shutdown cleanly.
1853 	 */
1854 	rp = NULL;
1855 	if (!load_and_check_logfile(vol, &rp) ||
1856 			!ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1857 		static const char *es1a = "Failed to load $LogFile";
1858 		static const char *es1b = "$LogFile is not clean";
1859 		static const char *es2 = ".  Mount in Windows.";
1860 		const char *es1;
1861 
1862 		es1 = !vol->logfile_ino ? es1a : es1b;
1863 		/* If a read-write mount, convert it to a read-only mount. */
1864 		if (!(sb->s_flags & MS_RDONLY)) {
1865 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1866 					ON_ERRORS_CONTINUE))) {
1867 				ntfs_error(sb, "%s and neither on_errors="
1868 						"continue nor on_errors="
1869 						"remount-ro was specified%s",
1870 						es1, es2);
1871 				if (vol->logfile_ino) {
1872 					BUG_ON(!rp);
1873 					ntfs_free(rp);
1874 				}
1875 				goto iput_logfile_err_out;
1876 			}
1877 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1878 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1879 		} else
1880 			ntfs_warning(sb, "%s.  Will not be able to remount "
1881 					"read-write%s", es1, es2);
1882 		/* This will prevent a read-write remount. */
1883 		NVolSetErrors(vol);
1884 	}
1885 	ntfs_free(rp);
1886 #endif /* NTFS_RW */
1887 	/* Get the root directory inode so we can do path lookups. */
1888 	vol->root_ino = ntfs_iget(sb, FILE_root);
1889 	if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1890 		if (!IS_ERR(vol->root_ino))
1891 			iput(vol->root_ino);
1892 		ntfs_error(sb, "Failed to load root directory.");
1893 		goto iput_logfile_err_out;
1894 	}
1895 #ifdef NTFS_RW
1896 	/*
1897 	 * Check if Windows is suspended to disk on the target volume.  If it
1898 	 * is hibernated, we must not write *anything* to the disk so set
1899 	 * NVolErrors() without setting the dirty volume flag and mount
1900 	 * read-only.  This will prevent read-write remounting and it will also
1901 	 * prevent all writes.
1902 	 */
1903 	err = check_windows_hibernation_status(vol);
1904 	if (unlikely(err)) {
1905 		static const char *es1a = "Failed to determine if Windows is "
1906 				"hibernated";
1907 		static const char *es1b = "Windows is hibernated";
1908 		static const char *es2 = ".  Run chkdsk.";
1909 		const char *es1;
1910 
1911 		es1 = err < 0 ? 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 				goto iput_root_err_out;
1921 			}
1922 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1923 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1924 		} else
1925 			ntfs_warning(sb, "%s.  Will not be able to remount "
1926 					"read-write%s", es1, es2);
1927 		/* This will prevent a read-write remount. */
1928 		NVolSetErrors(vol);
1929 	}
1930 	/* If (still) a read-write mount, mark the volume dirty. */
1931 	if (!(sb->s_flags & MS_RDONLY) &&
1932 			ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
1933 		static const char *es1 = "Failed to set dirty bit in volume "
1934 				"information flags";
1935 		static const char *es2 = ".  Run chkdsk.";
1936 
1937 		/* Convert to a read-only mount. */
1938 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1939 				ON_ERRORS_CONTINUE))) {
1940 			ntfs_error(sb, "%s and neither on_errors=continue nor "
1941 					"on_errors=remount-ro was specified%s",
1942 					es1, es2);
1943 			goto iput_root_err_out;
1944 		}
1945 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1946 		sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1947 		/*
1948 		 * Do not set NVolErrors() because ntfs_remount() might manage
1949 		 * to set the dirty flag in which case all would be well.
1950 		 */
1951 	}
1952 #if 0
1953 	// TODO: Enable this code once we start modifying anything that is
1954 	//	 different between NTFS 1.2 and 3.x...
1955 	/*
1956 	 * If (still) a read-write mount, set the NT4 compatibility flag on
1957 	 * newer NTFS version volumes.
1958 	 */
1959 	if (!(sb->s_flags & MS_RDONLY) && (vol->major_ver > 1) &&
1960 			ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
1961 		static const char *es1 = "Failed to set NT4 compatibility flag";
1962 		static const char *es2 = ".  Run chkdsk.";
1963 
1964 		/* Convert to a read-only mount. */
1965 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1966 				ON_ERRORS_CONTINUE))) {
1967 			ntfs_error(sb, "%s and neither on_errors=continue nor "
1968 					"on_errors=remount-ro was specified%s",
1969 					es1, es2);
1970 			goto iput_root_err_out;
1971 		}
1972 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1973 		sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1974 		NVolSetErrors(vol);
1975 	}
1976 #endif
1977 	/* If (still) a read-write mount, empty the logfile. */
1978 	if (!(sb->s_flags & MS_RDONLY) &&
1979 			!ntfs_empty_logfile(vol->logfile_ino)) {
1980 		static const char *es1 = "Failed to empty $LogFile";
1981 		static const char *es2 = ".  Mount in Windows.";
1982 
1983 		/* Convert to a read-only mount. */
1984 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1985 				ON_ERRORS_CONTINUE))) {
1986 			ntfs_error(sb, "%s and neither on_errors=continue nor "
1987 					"on_errors=remount-ro was specified%s",
1988 					es1, es2);
1989 			goto iput_root_err_out;
1990 		}
1991 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1992 		sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
1993 		NVolSetErrors(vol);
1994 	}
1995 #endif /* NTFS_RW */
1996 	/* If on NTFS versions before 3.0, we are done. */
1997 	if (unlikely(vol->major_ver < 3))
1998 		return TRUE;
1999 	/* NTFS 3.0+ specific initialization. */
2000 	/* Get the security descriptors inode. */
2001 	vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2002 	if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2003 		if (!IS_ERR(vol->secure_ino))
2004 			iput(vol->secure_ino);
2005 		ntfs_error(sb, "Failed to load $Secure.");
2006 		goto iput_root_err_out;
2007 	}
2008 	// TODO: Initialize security.
2009 	/* Get the extended system files' directory inode. */
2010 	vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2011 	if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2012 		if (!IS_ERR(vol->extend_ino))
2013 			iput(vol->extend_ino);
2014 		ntfs_error(sb, "Failed to load $Extend.");
2015 		goto iput_sec_err_out;
2016 	}
2017 #ifdef NTFS_RW
2018 	/* Find the quota file, load it if present, and set it up. */
2019 	if (!load_and_init_quota(vol)) {
2020 		static const char *es1 = "Failed to load $Quota";
2021 		static const char *es2 = ".  Run chkdsk.";
2022 
2023 		/* If a read-write mount, convert it to a read-only mount. */
2024 		if (!(sb->s_flags & MS_RDONLY)) {
2025 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2026 					ON_ERRORS_CONTINUE))) {
2027 				ntfs_error(sb, "%s and neither on_errors="
2028 						"continue nor on_errors="
2029 						"remount-ro was specified%s",
2030 						es1, es2);
2031 				goto iput_quota_err_out;
2032 			}
2033 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
2034 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2035 		} else
2036 			ntfs_warning(sb, "%s.  Will not be able to remount "
2037 					"read-write%s", es1, es2);
2038 		/* This will prevent a read-write remount. */
2039 		NVolSetErrors(vol);
2040 	}
2041 	/* If (still) a read-write mount, mark the quotas out of date. */
2042 	if (!(sb->s_flags & MS_RDONLY) &&
2043 			!ntfs_mark_quotas_out_of_date(vol)) {
2044 		static const char *es1 = "Failed to mark quotas out of date";
2045 		static const char *es2 = ".  Run chkdsk.";
2046 
2047 		/* Convert to a read-only mount. */
2048 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2049 				ON_ERRORS_CONTINUE))) {
2050 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2051 					"on_errors=remount-ro was specified%s",
2052 					es1, es2);
2053 			goto iput_quota_err_out;
2054 		}
2055 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2056 		sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
2057 		NVolSetErrors(vol);
2058 	}
2059 	/*
2060 	 * Find the transaction log file ($UsnJrnl), load it if present, check
2061 	 * it, and set it up.
2062 	 */
2063 	if (!load_and_init_usnjrnl(vol)) {
2064 		static const char *es1 = "Failed to load $UsnJrnl";
2065 		static const char *es2 = ".  Run chkdsk.";
2066 
2067 		/* If a read-write mount, convert it to a read-only mount. */
2068 		if (!(sb->s_flags & MS_RDONLY)) {
2069 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2070 					ON_ERRORS_CONTINUE))) {
2071 				ntfs_error(sb, "%s and neither on_errors="
2072 						"continue nor on_errors="
2073 						"remount-ro was specified%s",
2074 						es1, es2);
2075 				goto iput_usnjrnl_err_out;
2076 			}
2077 			sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
2078 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2079 		} else
2080 			ntfs_warning(sb, "%s.  Will not be able to remount "
2081 					"read-write%s", es1, es2);
2082 		/* This will prevent a read-write remount. */
2083 		NVolSetErrors(vol);
2084 	}
2085 	/* If (still) a read-write mount, stamp the transaction log. */
2086 	if (!(sb->s_flags & MS_RDONLY) && !ntfs_stamp_usnjrnl(vol)) {
2087 		static const char *es1 = "Failed to stamp transaction log "
2088 				"($UsnJrnl)";
2089 		static const char *es2 = ".  Run chkdsk.";
2090 
2091 		/* Convert to a read-only mount. */
2092 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2093 				ON_ERRORS_CONTINUE))) {
2094 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2095 					"on_errors=remount-ro was specified%s",
2096 					es1, es2);
2097 			goto iput_usnjrnl_err_out;
2098 		}
2099 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2100 		sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
2101 		NVolSetErrors(vol);
2102 	}
2103 #endif /* NTFS_RW */
2104 	return TRUE;
2105 #ifdef NTFS_RW
2106 iput_usnjrnl_err_out:
2107 	if (vol->usnjrnl_j_ino)
2108 		iput(vol->usnjrnl_j_ino);
2109 	if (vol->usnjrnl_max_ino)
2110 		iput(vol->usnjrnl_max_ino);
2111 	if (vol->usnjrnl_ino)
2112 		iput(vol->usnjrnl_ino);
2113 iput_quota_err_out:
2114 	if (vol->quota_q_ino)
2115 		iput(vol->quota_q_ino);
2116 	if (vol->quota_ino)
2117 		iput(vol->quota_ino);
2118 	iput(vol->extend_ino);
2119 #endif /* NTFS_RW */
2120 iput_sec_err_out:
2121 	iput(vol->secure_ino);
2122 iput_root_err_out:
2123 	iput(vol->root_ino);
2124 iput_logfile_err_out:
2125 #ifdef NTFS_RW
2126 	if (vol->logfile_ino)
2127 		iput(vol->logfile_ino);
2128 iput_vol_err_out:
2129 #endif /* NTFS_RW */
2130 	iput(vol->vol_ino);
2131 iput_lcnbmp_err_out:
2132 	iput(vol->lcnbmp_ino);
2133 iput_attrdef_err_out:
2134 	vol->attrdef_size = 0;
2135 	if (vol->attrdef) {
2136 		ntfs_free(vol->attrdef);
2137 		vol->attrdef = NULL;
2138 	}
2139 #ifdef NTFS_RW
2140 iput_upcase_err_out:
2141 #endif /* NTFS_RW */
2142 	vol->upcase_len = 0;
2143 	down(&ntfs_lock);
2144 	if (vol->upcase == default_upcase) {
2145 		ntfs_nr_upcase_users--;
2146 		vol->upcase = NULL;
2147 	}
2148 	up(&ntfs_lock);
2149 	if (vol->upcase) {
2150 		ntfs_free(vol->upcase);
2151 		vol->upcase = NULL;
2152 	}
2153 iput_mftbmp_err_out:
2154 	iput(vol->mftbmp_ino);
2155 iput_mirr_err_out:
2156 #ifdef NTFS_RW
2157 	if (vol->mftmirr_ino)
2158 		iput(vol->mftmirr_ino);
2159 #endif /* NTFS_RW */
2160 	return FALSE;
2161 }
2162 
2163 /**
2164  * ntfs_put_super - called by the vfs to unmount a volume
2165  * @sb:		vfs superblock of volume to unmount
2166  *
2167  * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2168  * the volume is being unmounted (umount system call has been invoked) and it
2169  * releases all inodes and memory belonging to the NTFS specific part of the
2170  * super block.
2171  */
2172 static void ntfs_put_super(struct super_block *sb)
2173 {
2174 	ntfs_volume *vol = NTFS_SB(sb);
2175 
2176 	ntfs_debug("Entering.");
2177 #ifdef NTFS_RW
2178 	/*
2179 	 * Commit all inodes while they are still open in case some of them
2180 	 * cause others to be dirtied.
2181 	 */
2182 	ntfs_commit_inode(vol->vol_ino);
2183 
2184 	/* NTFS 3.0+ specific. */
2185 	if (vol->major_ver >= 3) {
2186 		if (vol->usnjrnl_j_ino)
2187 			ntfs_commit_inode(vol->usnjrnl_j_ino);
2188 		if (vol->usnjrnl_max_ino)
2189 			ntfs_commit_inode(vol->usnjrnl_max_ino);
2190 		if (vol->usnjrnl_ino)
2191 			ntfs_commit_inode(vol->usnjrnl_ino);
2192 		if (vol->quota_q_ino)
2193 			ntfs_commit_inode(vol->quota_q_ino);
2194 		if (vol->quota_ino)
2195 			ntfs_commit_inode(vol->quota_ino);
2196 		if (vol->extend_ino)
2197 			ntfs_commit_inode(vol->extend_ino);
2198 		if (vol->secure_ino)
2199 			ntfs_commit_inode(vol->secure_ino);
2200 	}
2201 
2202 	ntfs_commit_inode(vol->root_ino);
2203 
2204 	down_write(&vol->lcnbmp_lock);
2205 	ntfs_commit_inode(vol->lcnbmp_ino);
2206 	up_write(&vol->lcnbmp_lock);
2207 
2208 	down_write(&vol->mftbmp_lock);
2209 	ntfs_commit_inode(vol->mftbmp_ino);
2210 	up_write(&vol->mftbmp_lock);
2211 
2212 	if (vol->logfile_ino)
2213 		ntfs_commit_inode(vol->logfile_ino);
2214 
2215 	if (vol->mftmirr_ino)
2216 		ntfs_commit_inode(vol->mftmirr_ino);
2217 	ntfs_commit_inode(vol->mft_ino);
2218 
2219 	/*
2220 	 * If a read-write mount and no volume errors have occured, mark the
2221 	 * volume clean.  Also, re-commit all affected inodes.
2222 	 */
2223 	if (!(sb->s_flags & MS_RDONLY)) {
2224 		if (!NVolErrors(vol)) {
2225 			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2226 				ntfs_warning(sb, "Failed to clear dirty bit "
2227 						"in volume information "
2228 						"flags.  Run chkdsk.");
2229 			ntfs_commit_inode(vol->vol_ino);
2230 			ntfs_commit_inode(vol->root_ino);
2231 			if (vol->mftmirr_ino)
2232 				ntfs_commit_inode(vol->mftmirr_ino);
2233 			ntfs_commit_inode(vol->mft_ino);
2234 		} else {
2235 			ntfs_warning(sb, "Volume has errors.  Leaving volume "
2236 					"marked dirty.  Run chkdsk.");
2237 		}
2238 	}
2239 #endif /* NTFS_RW */
2240 
2241 	iput(vol->vol_ino);
2242 	vol->vol_ino = NULL;
2243 
2244 	/* NTFS 3.0+ specific clean up. */
2245 	if (vol->major_ver >= 3) {
2246 #ifdef NTFS_RW
2247 		if (vol->usnjrnl_j_ino) {
2248 			iput(vol->usnjrnl_j_ino);
2249 			vol->usnjrnl_j_ino = NULL;
2250 		}
2251 		if (vol->usnjrnl_max_ino) {
2252 			iput(vol->usnjrnl_max_ino);
2253 			vol->usnjrnl_max_ino = NULL;
2254 		}
2255 		if (vol->usnjrnl_ino) {
2256 			iput(vol->usnjrnl_ino);
2257 			vol->usnjrnl_ino = NULL;
2258 		}
2259 		if (vol->quota_q_ino) {
2260 			iput(vol->quota_q_ino);
2261 			vol->quota_q_ino = NULL;
2262 		}
2263 		if (vol->quota_ino) {
2264 			iput(vol->quota_ino);
2265 			vol->quota_ino = NULL;
2266 		}
2267 #endif /* NTFS_RW */
2268 		if (vol->extend_ino) {
2269 			iput(vol->extend_ino);
2270 			vol->extend_ino = NULL;
2271 		}
2272 		if (vol->secure_ino) {
2273 			iput(vol->secure_ino);
2274 			vol->secure_ino = NULL;
2275 		}
2276 	}
2277 
2278 	iput(vol->root_ino);
2279 	vol->root_ino = NULL;
2280 
2281 	down_write(&vol->lcnbmp_lock);
2282 	iput(vol->lcnbmp_ino);
2283 	vol->lcnbmp_ino = NULL;
2284 	up_write(&vol->lcnbmp_lock);
2285 
2286 	down_write(&vol->mftbmp_lock);
2287 	iput(vol->mftbmp_ino);
2288 	vol->mftbmp_ino = NULL;
2289 	up_write(&vol->mftbmp_lock);
2290 
2291 #ifdef NTFS_RW
2292 	if (vol->logfile_ino) {
2293 		iput(vol->logfile_ino);
2294 		vol->logfile_ino = NULL;
2295 	}
2296 	if (vol->mftmirr_ino) {
2297 		/* Re-commit the mft mirror and mft just in case. */
2298 		ntfs_commit_inode(vol->mftmirr_ino);
2299 		ntfs_commit_inode(vol->mft_ino);
2300 		iput(vol->mftmirr_ino);
2301 		vol->mftmirr_ino = NULL;
2302 	}
2303 	/*
2304 	 * If any dirty inodes are left, throw away all mft data page cache
2305 	 * pages to allow a clean umount.  This should never happen any more
2306 	 * due to mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2307 	 * the underlying mft records are written out and cleaned.  If it does,
2308 	 * happen anyway, we want to know...
2309 	 */
2310 	ntfs_commit_inode(vol->mft_ino);
2311 	write_inode_now(vol->mft_ino, 1);
2312 	if (!list_empty(&sb->s_dirty)) {
2313 		const char *s1, *s2;
2314 
2315 		down(&vol->mft_ino->i_sem);
2316 		truncate_inode_pages(vol->mft_ino->i_mapping, 0);
2317 		up(&vol->mft_ino->i_sem);
2318 		write_inode_now(vol->mft_ino, 1);
2319 		if (!list_empty(&sb->s_dirty)) {
2320 			static const char *_s1 = "inodes";
2321 			static const char *_s2 = "";
2322 			s1 = _s1;
2323 			s2 = _s2;
2324 		} else {
2325 			static const char *_s1 = "mft pages";
2326 			static const char *_s2 = "They have been thrown "
2327 					"away.  ";
2328 			s1 = _s1;
2329 			s2 = _s2;
2330 		}
2331 		ntfs_error(sb, "Dirty %s found at umount time.  %sYou should "
2332 				"run chkdsk.  Please email "
2333 				"linux-ntfs-dev@lists.sourceforge.net and say "
2334 				"that you saw this message.  Thank you.", s1,
2335 				s2);
2336 	}
2337 #endif /* NTFS_RW */
2338 
2339 	iput(vol->mft_ino);
2340 	vol->mft_ino = NULL;
2341 
2342 	/* Throw away the table of attribute definitions. */
2343 	vol->attrdef_size = 0;
2344 	if (vol->attrdef) {
2345 		ntfs_free(vol->attrdef);
2346 		vol->attrdef = NULL;
2347 	}
2348 	vol->upcase_len = 0;
2349 	/*
2350 	 * Destroy the global default upcase table if necessary.  Also decrease
2351 	 * the number of upcase users if we are a user.
2352 	 */
2353 	down(&ntfs_lock);
2354 	if (vol->upcase == default_upcase) {
2355 		ntfs_nr_upcase_users--;
2356 		vol->upcase = NULL;
2357 	}
2358 	if (!ntfs_nr_upcase_users && default_upcase) {
2359 		ntfs_free(default_upcase);
2360 		default_upcase = NULL;
2361 	}
2362 	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2363 		free_compression_buffers();
2364 	up(&ntfs_lock);
2365 	if (vol->upcase) {
2366 		ntfs_free(vol->upcase);
2367 		vol->upcase = NULL;
2368 	}
2369 	if (vol->nls_map) {
2370 		unload_nls(vol->nls_map);
2371 		vol->nls_map = NULL;
2372 	}
2373 	sb->s_fs_info = NULL;
2374 	kfree(vol);
2375 	return;
2376 }
2377 
2378 /**
2379  * get_nr_free_clusters - return the number of free clusters on a volume
2380  * @vol:	ntfs volume for which to obtain free cluster count
2381  *
2382  * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2383  * actually calculate the number of clusters in use instead because this
2384  * allows us to not care about partial pages as these will be just zero filled
2385  * and hence not be counted as allocated clusters.
2386  *
2387  * The only particularity is that clusters beyond the end of the logical ntfs
2388  * volume will be marked as allocated to prevent errors which means we have to
2389  * discount those at the end. This is important as the cluster bitmap always
2390  * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2391  * the logical volume and marked in use when they are not as they do not exist.
2392  *
2393  * If any pages cannot be read we assume all clusters in the erroring pages are
2394  * in use. This means we return an underestimate on errors which is better than
2395  * an overestimate.
2396  */
2397 static s64 get_nr_free_clusters(ntfs_volume *vol)
2398 {
2399 	s64 nr_free = vol->nr_clusters;
2400 	u32 *kaddr;
2401 	struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2402 	filler_t *readpage = (filler_t*)mapping->a_ops->readpage;
2403 	struct page *page;
2404 	pgoff_t index, max_index;
2405 
2406 	ntfs_debug("Entering.");
2407 	/* Serialize accesses to the cluster bitmap. */
2408 	down_read(&vol->lcnbmp_lock);
2409 	/*
2410 	 * Convert the number of bits into bytes rounded up, then convert into
2411 	 * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2412 	 * full and one partial page max_index = 2.
2413 	 */
2414 	max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2415 			PAGE_CACHE_SHIFT;
2416 	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2417 	ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2418 			max_index, PAGE_CACHE_SIZE / 4);
2419 	for (index = 0; index < max_index; index++) {
2420 		unsigned int i;
2421 		/*
2422 		 * Read the page from page cache, getting it from backing store
2423 		 * if necessary, and increment the use count.
2424 		 */
2425 		page = read_cache_page(mapping, index, (filler_t*)readpage,
2426 				NULL);
2427 		/* Ignore pages which errored synchronously. */
2428 		if (IS_ERR(page)) {
2429 			ntfs_debug("Sync read_cache_page() error. Skipping "
2430 					"page (index 0x%lx).", index);
2431 			nr_free -= PAGE_CACHE_SIZE * 8;
2432 			continue;
2433 		}
2434 		wait_on_page_locked(page);
2435 		/* Ignore pages which errored asynchronously. */
2436 		if (!PageUptodate(page)) {
2437 			ntfs_debug("Async read_cache_page() error. Skipping "
2438 					"page (index 0x%lx).", index);
2439 			page_cache_release(page);
2440 			nr_free -= PAGE_CACHE_SIZE * 8;
2441 			continue;
2442 		}
2443 		kaddr = (u32*)kmap_atomic(page, KM_USER0);
2444 		/*
2445 		 * For each 4 bytes, subtract the number of set bits. If this
2446 		 * is the last page and it is partial we don't really care as
2447 		 * it just means we do a little extra work but it won't affect
2448 		 * the result as all out of range bytes are set to zero by
2449 		 * ntfs_readpage().
2450 		 */
2451 	  	for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2452 			nr_free -= (s64)hweight32(kaddr[i]);
2453 		kunmap_atomic(kaddr, KM_USER0);
2454 		page_cache_release(page);
2455 	}
2456 	ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2457 	/*
2458 	 * Fixup for eventual bits outside logical ntfs volume (see function
2459 	 * description above).
2460 	 */
2461 	if (vol->nr_clusters & 63)
2462 		nr_free += 64 - (vol->nr_clusters & 63);
2463 	up_read(&vol->lcnbmp_lock);
2464 	/* If errors occured we may well have gone below zero, fix this. */
2465 	if (nr_free < 0)
2466 		nr_free = 0;
2467 	ntfs_debug("Exiting.");
2468 	return nr_free;
2469 }
2470 
2471 /**
2472  * __get_nr_free_mft_records - return the number of free inodes on a volume
2473  * @vol:	ntfs volume for which to obtain free inode count
2474  * @nr_free:	number of mft records in filesystem
2475  * @max_index:	maximum number of pages containing set bits
2476  *
2477  * Calculate the number of free mft records (inodes) on the mounted NTFS
2478  * volume @vol. We actually calculate the number of mft records in use instead
2479  * because this allows us to not care about partial pages as these will be just
2480  * zero filled and hence not be counted as allocated mft record.
2481  *
2482  * If any pages cannot be read we assume all mft records in the erroring pages
2483  * are in use. This means we return an underestimate on errors which is better
2484  * than an overestimate.
2485  *
2486  * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2487  */
2488 static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2489 		s64 nr_free, const pgoff_t max_index)
2490 {
2491 	u32 *kaddr;
2492 	struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2493 	filler_t *readpage = (filler_t*)mapping->a_ops->readpage;
2494 	struct page *page;
2495 	pgoff_t index;
2496 
2497 	ntfs_debug("Entering.");
2498 	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2499 	ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2500 			"0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2501 	for (index = 0; index < max_index; index++) {
2502 		unsigned int i;
2503 		/*
2504 		 * Read the page from page cache, getting it from backing store
2505 		 * if necessary, and increment the use count.
2506 		 */
2507 		page = read_cache_page(mapping, index, (filler_t*)readpage,
2508 				NULL);
2509 		/* Ignore pages which errored synchronously. */
2510 		if (IS_ERR(page)) {
2511 			ntfs_debug("Sync read_cache_page() error. Skipping "
2512 					"page (index 0x%lx).", index);
2513 			nr_free -= PAGE_CACHE_SIZE * 8;
2514 			continue;
2515 		}
2516 		wait_on_page_locked(page);
2517 		/* Ignore pages which errored asynchronously. */
2518 		if (!PageUptodate(page)) {
2519 			ntfs_debug("Async read_cache_page() error. Skipping "
2520 					"page (index 0x%lx).", index);
2521 			page_cache_release(page);
2522 			nr_free -= PAGE_CACHE_SIZE * 8;
2523 			continue;
2524 		}
2525 		kaddr = (u32*)kmap_atomic(page, KM_USER0);
2526 		/*
2527 		 * For each 4 bytes, subtract the number of set bits. If this
2528 		 * is the last page and it is partial we don't really care as
2529 		 * it just means we do a little extra work but it won't affect
2530 		 * the result as all out of range bytes are set to zero by
2531 		 * ntfs_readpage().
2532 		 */
2533 	  	for (i = 0; i < PAGE_CACHE_SIZE / 4; i++)
2534 			nr_free -= (s64)hweight32(kaddr[i]);
2535 		kunmap_atomic(kaddr, KM_USER0);
2536 		page_cache_release(page);
2537 	}
2538 	ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2539 			index - 1);
2540 	/* If errors occured we may well have gone below zero, fix this. */
2541 	if (nr_free < 0)
2542 		nr_free = 0;
2543 	ntfs_debug("Exiting.");
2544 	return nr_free;
2545 }
2546 
2547 /**
2548  * ntfs_statfs - return information about mounted NTFS volume
2549  * @sb:		super block of mounted volume
2550  * @sfs:	statfs structure in which to return the information
2551  *
2552  * Return information about the mounted NTFS volume @sb in the statfs structure
2553  * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2554  * called). We interpret the values to be correct of the moment in time at
2555  * which we are called. Most values are variable otherwise and this isn't just
2556  * the free values but the totals as well. For example we can increase the
2557  * total number of file nodes if we run out and we can keep doing this until
2558  * there is no more space on the volume left at all.
2559  *
2560  * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2561  * ustat system calls.
2562  *
2563  * Return 0 on success or -errno on error.
2564  */
2565 static int ntfs_statfs(struct super_block *sb, struct kstatfs *sfs)
2566 {
2567 	s64 size;
2568 	ntfs_volume *vol = NTFS_SB(sb);
2569 	ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2570 	pgoff_t max_index;
2571 	unsigned long flags;
2572 
2573 	ntfs_debug("Entering.");
2574 	/* Type of filesystem. */
2575 	sfs->f_type   = NTFS_SB_MAGIC;
2576 	/* Optimal transfer block size. */
2577 	sfs->f_bsize  = PAGE_CACHE_SIZE;
2578 	/*
2579 	 * Total data blocks in filesystem in units of f_bsize and since
2580 	 * inodes are also stored in data blocs ($MFT is a file) this is just
2581 	 * the total clusters.
2582 	 */
2583 	sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2584 				PAGE_CACHE_SHIFT;
2585 	/* Free data blocks in filesystem in units of f_bsize. */
2586 	size	      = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2587 				PAGE_CACHE_SHIFT;
2588 	if (size < 0LL)
2589 		size = 0LL;
2590 	/* Free blocks avail to non-superuser, same as above on NTFS. */
2591 	sfs->f_bavail = sfs->f_bfree = size;
2592 	/* Serialize accesses to the inode bitmap. */
2593 	down_read(&vol->mftbmp_lock);
2594 	read_lock_irqsave(&mft_ni->size_lock, flags);
2595 	size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2596 	/*
2597 	 * Convert the maximum number of set bits into bytes rounded up, then
2598 	 * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2599 	 * have one full and one partial page max_index = 2.
2600 	 */
2601 	max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2602 			+ 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2603 	read_unlock_irqrestore(&mft_ni->size_lock, flags);
2604 	/* Number of inodes in filesystem (at this point in time). */
2605 	sfs->f_files = size;
2606 	/* Free inodes in fs (based on current total count). */
2607 	sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2608 	up_read(&vol->mftbmp_lock);
2609 	/*
2610 	 * File system id. This is extremely *nix flavour dependent and even
2611 	 * within Linux itself all fs do their own thing. I interpret this to
2612 	 * mean a unique id associated with the mounted fs and not the id
2613 	 * associated with the filesystem driver, the latter is already given
2614 	 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2615 	 * volume serial number splitting it into two 32-bit parts. We enter
2616 	 * the least significant 32-bits in f_fsid[0] and the most significant
2617 	 * 32-bits in f_fsid[1].
2618 	 */
2619 	sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2620 	sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2621 	/* Maximum length of filenames. */
2622 	sfs->f_namelen	   = NTFS_MAX_NAME_LEN;
2623 	return 0;
2624 }
2625 
2626 /**
2627  * The complete super operations.
2628  */
2629 static struct super_operations ntfs_sops = {
2630 	.alloc_inode	= ntfs_alloc_big_inode,	  /* VFS: Allocate new inode. */
2631 	.destroy_inode	= ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2632 	.put_inode	= ntfs_put_inode,	  /* VFS: Called just before
2633 						     the inode reference count
2634 						     is decreased. */
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  */
2683 static 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 result;
2689 
2690 	ntfs_debug("Entering.");
2691 #ifndef NTFS_RW
2692 	sb->s_flags |= MS_RDONLY | MS_NOATIME | MS_NODIRATIME;
2693 #endif /* ! NTFS_RW */
2694 	/* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2695 	sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2696 	vol = NTFS_SB(sb);
2697 	if (!vol) {
2698 		if (!silent)
2699 			ntfs_error(sb, "Allocation of NTFS volume structure "
2700 					"failed. Aborting mount...");
2701 		return -ENOMEM;
2702 	}
2703 	/* Initialize ntfs_volume structure. */
2704 	*vol = (ntfs_volume) {
2705 		.sb = sb,
2706 		/*
2707 		 * Default is group and other don't have any access to files or
2708 		 * directories while owner has full access. Further, files by
2709 		 * default are not executable but directories are of course
2710 		 * browseable.
2711 		 */
2712 		.fmask = 0177,
2713 		.dmask = 0077,
2714 	};
2715 	init_rwsem(&vol->mftbmp_lock);
2716 	init_rwsem(&vol->lcnbmp_lock);
2717 
2718 	unlock_kernel();
2719 
2720 	/* By default, enable sparse support. */
2721 	NVolSetSparseEnabled(vol);
2722 
2723 	/* Important to get the mount options dealt with now. */
2724 	if (!parse_options(vol, (char*)opt))
2725 		goto err_out_now;
2726 
2727 	/*
2728 	 * TODO: Fail safety check. In the future we should really be able to
2729 	 * cope with this being the case, but for now just bail out.
2730 	 */
2731 	if (bdev_hardsect_size(sb->s_bdev) > NTFS_BLOCK_SIZE) {
2732 		if (!silent)
2733 			ntfs_error(sb, "Device has unsupported hardsect_size.");
2734 		goto err_out_now;
2735 	}
2736 
2737 	/* Setup the device access block size to NTFS_BLOCK_SIZE. */
2738 	if (sb_set_blocksize(sb, NTFS_BLOCK_SIZE) != NTFS_BLOCK_SIZE) {
2739 		if (!silent)
2740 			ntfs_error(sb, "Unable to set block size.");
2741 		goto err_out_now;
2742 	}
2743 
2744 	/* Get the size of the device in units of NTFS_BLOCK_SIZE bytes. */
2745 	vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2746 			NTFS_BLOCK_SIZE_BITS;
2747 
2748 	/* Read the boot sector and return unlocked buffer head to it. */
2749 	if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2750 		if (!silent)
2751 			ntfs_error(sb, "Not an NTFS volume.");
2752 		goto err_out_now;
2753 	}
2754 
2755 	/*
2756 	 * Extract the data from the boot sector and setup the ntfs super block
2757 	 * using it.
2758 	 */
2759 	result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2760 
2761 	/* Initialize the cluster and mft allocators. */
2762 	ntfs_setup_allocators(vol);
2763 
2764 	brelse(bh);
2765 
2766 	if (!result) {
2767 		if (!silent)
2768 			ntfs_error(sb, "Unsupported NTFS filesystem.");
2769 		goto err_out_now;
2770 	}
2771 
2772 	/*
2773 	 * TODO: When we start coping with sector sizes different from
2774 	 * NTFS_BLOCK_SIZE, we now probably need to set the blocksize of the
2775 	 * device (probably to NTFS_BLOCK_SIZE).
2776 	 */
2777 
2778 	/* Setup remaining fields in the super block. */
2779 	sb->s_magic = NTFS_SB_MAGIC;
2780 
2781 	/*
2782 	 * Ntfs allows 63 bits for the file size, i.e. correct would be:
2783 	 *	sb->s_maxbytes = ~0ULL >> 1;
2784 	 * But the kernel uses a long as the page cache page index which on
2785 	 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2786 	 * defined to the maximum the page cache page index can cope with
2787 	 * without overflowing the index or to 2^63 - 1, whichever is smaller.
2788 	 */
2789 	sb->s_maxbytes = MAX_LFS_FILESIZE;
2790 
2791 	sb->s_time_gran = 100;
2792 
2793 	/*
2794 	 * Now load the metadata required for the page cache and our address
2795 	 * space operations to function. We do this by setting up a specialised
2796 	 * read_inode method and then just calling the normal iget() to obtain
2797 	 * the inode for $MFT which is sufficient to allow our normal inode
2798 	 * operations and associated address space operations to function.
2799 	 */
2800 	sb->s_op = &ntfs_sops;
2801 	tmp_ino = new_inode(sb);
2802 	if (!tmp_ino) {
2803 		if (!silent)
2804 			ntfs_error(sb, "Failed to load essential metadata.");
2805 		goto err_out_now;
2806 	}
2807 	tmp_ino->i_ino = FILE_MFT;
2808 	insert_inode_hash(tmp_ino);
2809 	if (ntfs_read_inode_mount(tmp_ino) < 0) {
2810 		if (!silent)
2811 			ntfs_error(sb, "Failed to load essential metadata.");
2812 		goto iput_tmp_ino_err_out_now;
2813 	}
2814 	down(&ntfs_lock);
2815 	/*
2816 	 * The current mount is a compression user if the cluster size is
2817 	 * less than or equal 4kiB.
2818 	 */
2819 	if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2820 		result = allocate_compression_buffers();
2821 		if (result) {
2822 			ntfs_error(NULL, "Failed to allocate buffers "
2823 					"for compression engine.");
2824 			ntfs_nr_compression_users--;
2825 			up(&ntfs_lock);
2826 			goto iput_tmp_ino_err_out_now;
2827 		}
2828 	}
2829 	/*
2830 	 * Generate the global default upcase table if necessary.  Also
2831 	 * temporarily increment the number of upcase users to avoid race
2832 	 * conditions with concurrent (u)mounts.
2833 	 */
2834 	if (!default_upcase)
2835 		default_upcase = generate_default_upcase();
2836 	ntfs_nr_upcase_users++;
2837 	up(&ntfs_lock);
2838 	/*
2839 	 * From now on, ignore @silent parameter. If we fail below this line,
2840 	 * it will be due to a corrupt fs or a system error, so we report it.
2841 	 */
2842 	/*
2843 	 * Open the system files with normal access functions and complete
2844 	 * setting up the ntfs super block.
2845 	 */
2846 	if (!load_system_files(vol)) {
2847 		ntfs_error(sb, "Failed to load system files.");
2848 		goto unl_upcase_iput_tmp_ino_err_out_now;
2849 	}
2850 	if ((sb->s_root = d_alloc_root(vol->root_ino))) {
2851 		/* We increment i_count simulating an ntfs_iget(). */
2852 		atomic_inc(&vol->root_ino->i_count);
2853 		ntfs_debug("Exiting, status successful.");
2854 		/* Release the default upcase if it has no users. */
2855 		down(&ntfs_lock);
2856 		if (!--ntfs_nr_upcase_users && default_upcase) {
2857 			ntfs_free(default_upcase);
2858 			default_upcase = NULL;
2859 		}
2860 		up(&ntfs_lock);
2861 		sb->s_export_op = &ntfs_export_ops;
2862 		lock_kernel();
2863 		return 0;
2864 	}
2865 	ntfs_error(sb, "Failed to allocate root directory.");
2866 	/* Clean up after the successful load_system_files() call from above. */
2867 	// TODO: Use ntfs_put_super() instead of repeating all this code...
2868 	// FIXME: Should mark the volume clean as the error is most likely
2869 	// 	  -ENOMEM.
2870 	iput(vol->vol_ino);
2871 	vol->vol_ino = NULL;
2872 	/* NTFS 3.0+ specific clean up. */
2873 	if (vol->major_ver >= 3) {
2874 #ifdef NTFS_RW
2875 		if (vol->usnjrnl_j_ino) {
2876 			iput(vol->usnjrnl_j_ino);
2877 			vol->usnjrnl_j_ino = NULL;
2878 		}
2879 		if (vol->usnjrnl_max_ino) {
2880 			iput(vol->usnjrnl_max_ino);
2881 			vol->usnjrnl_max_ino = NULL;
2882 		}
2883 		if (vol->usnjrnl_ino) {
2884 			iput(vol->usnjrnl_ino);
2885 			vol->usnjrnl_ino = NULL;
2886 		}
2887 		if (vol->quota_q_ino) {
2888 			iput(vol->quota_q_ino);
2889 			vol->quota_q_ino = NULL;
2890 		}
2891 		if (vol->quota_ino) {
2892 			iput(vol->quota_ino);
2893 			vol->quota_ino = NULL;
2894 		}
2895 #endif /* NTFS_RW */
2896 		if (vol->extend_ino) {
2897 			iput(vol->extend_ino);
2898 			vol->extend_ino = NULL;
2899 		}
2900 		if (vol->secure_ino) {
2901 			iput(vol->secure_ino);
2902 			vol->secure_ino = NULL;
2903 		}
2904 	}
2905 	iput(vol->root_ino);
2906 	vol->root_ino = NULL;
2907 	iput(vol->lcnbmp_ino);
2908 	vol->lcnbmp_ino = NULL;
2909 	iput(vol->mftbmp_ino);
2910 	vol->mftbmp_ino = NULL;
2911 #ifdef NTFS_RW
2912 	if (vol->logfile_ino) {
2913 		iput(vol->logfile_ino);
2914 		vol->logfile_ino = NULL;
2915 	}
2916 	if (vol->mftmirr_ino) {
2917 		iput(vol->mftmirr_ino);
2918 		vol->mftmirr_ino = NULL;
2919 	}
2920 #endif /* NTFS_RW */
2921 	/* Throw away the table of attribute definitions. */
2922 	vol->attrdef_size = 0;
2923 	if (vol->attrdef) {
2924 		ntfs_free(vol->attrdef);
2925 		vol->attrdef = NULL;
2926 	}
2927 	vol->upcase_len = 0;
2928 	down(&ntfs_lock);
2929 	if (vol->upcase == default_upcase) {
2930 		ntfs_nr_upcase_users--;
2931 		vol->upcase = NULL;
2932 	}
2933 	up(&ntfs_lock);
2934 	if (vol->upcase) {
2935 		ntfs_free(vol->upcase);
2936 		vol->upcase = NULL;
2937 	}
2938 	if (vol->nls_map) {
2939 		unload_nls(vol->nls_map);
2940 		vol->nls_map = NULL;
2941 	}
2942 	/* Error exit code path. */
2943 unl_upcase_iput_tmp_ino_err_out_now:
2944 	/*
2945 	 * Decrease the number of upcase users and destroy the global default
2946 	 * upcase table if necessary.
2947 	 */
2948 	down(&ntfs_lock);
2949 	if (!--ntfs_nr_upcase_users && default_upcase) {
2950 		ntfs_free(default_upcase);
2951 		default_upcase = NULL;
2952 	}
2953 	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2954 		free_compression_buffers();
2955 	up(&ntfs_lock);
2956 iput_tmp_ino_err_out_now:
2957 	iput(tmp_ino);
2958 	if (vol->mft_ino && vol->mft_ino != tmp_ino)
2959 		iput(vol->mft_ino);
2960 	vol->mft_ino = NULL;
2961 	/*
2962 	 * This is needed to get ntfs_clear_extent_inode() called for each
2963 	 * inode we have ever called ntfs_iget()/iput() on, otherwise we A)
2964 	 * leak resources and B) a subsequent mount fails automatically due to
2965 	 * ntfs_iget() never calling down into our ntfs_read_locked_inode()
2966 	 * method again... FIXME: Do we need to do this twice now because of
2967 	 * attribute inodes? I think not, so leave as is for now... (AIA)
2968 	 */
2969 	if (invalidate_inodes(sb)) {
2970 		ntfs_error(sb, "Busy inodes left. This is most likely a NTFS "
2971 				"driver bug.");
2972 		/* Copied from fs/super.c. I just love this message. (-; */
2973 		printk("NTFS: Busy inodes after umount. Self-destruct in 5 "
2974 				"seconds.  Have a nice day...\n");
2975 	}
2976 	/* Errors at this stage are irrelevant. */
2977 err_out_now:
2978 	lock_kernel();
2979 	sb->s_fs_info = NULL;
2980 	kfree(vol);
2981 	ntfs_debug("Failed, returning -EINVAL.");
2982 	return -EINVAL;
2983 }
2984 
2985 /*
2986  * This is a slab cache to optimize allocations and deallocations of Unicode
2987  * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
2988  * (255) Unicode characters + a terminating NULL Unicode character.
2989  */
2990 kmem_cache_t *ntfs_name_cache;
2991 
2992 /* Slab caches for efficient allocation/deallocation of inodes. */
2993 kmem_cache_t *ntfs_inode_cache;
2994 kmem_cache_t *ntfs_big_inode_cache;
2995 
2996 /* Init once constructor for the inode slab cache. */
2997 static void ntfs_big_inode_init_once(void *foo, kmem_cache_t *cachep,
2998 		unsigned long flags)
2999 {
3000 	ntfs_inode *ni = (ntfs_inode *)foo;
3001 
3002 	if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
3003 			SLAB_CTOR_CONSTRUCTOR)
3004 		inode_init_once(VFS_I(ni));
3005 }
3006 
3007 /*
3008  * Slab caches to optimize allocations and deallocations of attribute search
3009  * contexts and index contexts, respectively.
3010  */
3011 kmem_cache_t *ntfs_attr_ctx_cache;
3012 kmem_cache_t *ntfs_index_ctx_cache;
3013 
3014 /* Driver wide semaphore. */
3015 DECLARE_MUTEX(ntfs_lock);
3016 
3017 static struct super_block *ntfs_get_sb(struct file_system_type *fs_type,
3018 	int flags, const char *dev_name, void *data)
3019 {
3020 	return get_sb_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
3021 }
3022 
3023 static struct file_system_type ntfs_fs_type = {
3024 	.owner		= THIS_MODULE,
3025 	.name		= "ntfs",
3026 	.get_sb		= ntfs_get_sb,
3027 	.kill_sb	= kill_block_super,
3028 	.fs_flags	= FS_REQUIRES_DEV,
3029 };
3030 
3031 /* Stable names for the slab caches. */
3032 static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3033 static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3034 static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3035 static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3036 static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3037 
3038 static int __init init_ntfs_fs(void)
3039 {
3040 	int err = 0;
3041 
3042 	/* This may be ugly but it results in pretty output so who cares. (-8 */
3043 	printk(KERN_INFO "NTFS driver " NTFS_VERSION " [Flags: R/"
3044 #ifdef NTFS_RW
3045 			"W"
3046 #else
3047 			"O"
3048 #endif
3049 #ifdef DEBUG
3050 			" DEBUG"
3051 #endif
3052 #ifdef MODULE
3053 			" MODULE"
3054 #endif
3055 			"].\n");
3056 
3057 	ntfs_debug("Debug messages are enabled.");
3058 
3059 	ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3060 			sizeof(ntfs_index_context), 0 /* offset */,
3061 			SLAB_HWCACHE_ALIGN, NULL /* ctor */, NULL /* dtor */);
3062 	if (!ntfs_index_ctx_cache) {
3063 		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3064 				ntfs_index_ctx_cache_name);
3065 		goto ictx_err_out;
3066 	}
3067 	ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3068 			sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3069 			SLAB_HWCACHE_ALIGN, NULL /* ctor */, NULL /* dtor */);
3070 	if (!ntfs_attr_ctx_cache) {
3071 		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3072 				ntfs_attr_ctx_cache_name);
3073 		goto actx_err_out;
3074 	}
3075 
3076 	ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3077 			(NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3078 			SLAB_HWCACHE_ALIGN, NULL, NULL);
3079 	if (!ntfs_name_cache) {
3080 		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3081 				ntfs_name_cache_name);
3082 		goto name_err_out;
3083 	}
3084 
3085 	ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3086 			sizeof(ntfs_inode), 0,
3087 			SLAB_RECLAIM_ACCOUNT, NULL, NULL);
3088 	if (!ntfs_inode_cache) {
3089 		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3090 				ntfs_inode_cache_name);
3091 		goto inode_err_out;
3092 	}
3093 
3094 	ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3095 			sizeof(big_ntfs_inode), 0,
3096 			SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT,
3097 			ntfs_big_inode_init_once, NULL);
3098 	if (!ntfs_big_inode_cache) {
3099 		printk(KERN_CRIT "NTFS: Failed to create %s!\n",
3100 				ntfs_big_inode_cache_name);
3101 		goto big_inode_err_out;
3102 	}
3103 
3104 	/* Register the ntfs sysctls. */
3105 	err = ntfs_sysctl(1);
3106 	if (err) {
3107 		printk(KERN_CRIT "NTFS: Failed to register NTFS sysctls!\n");
3108 		goto sysctl_err_out;
3109 	}
3110 
3111 	err = register_filesystem(&ntfs_fs_type);
3112 	if (!err) {
3113 		ntfs_debug("NTFS driver registered successfully.");
3114 		return 0; /* Success! */
3115 	}
3116 	printk(KERN_CRIT "NTFS: Failed to register NTFS filesystem driver!\n");
3117 
3118 sysctl_err_out:
3119 	kmem_cache_destroy(ntfs_big_inode_cache);
3120 big_inode_err_out:
3121 	kmem_cache_destroy(ntfs_inode_cache);
3122 inode_err_out:
3123 	kmem_cache_destroy(ntfs_name_cache);
3124 name_err_out:
3125 	kmem_cache_destroy(ntfs_attr_ctx_cache);
3126 actx_err_out:
3127 	kmem_cache_destroy(ntfs_index_ctx_cache);
3128 ictx_err_out:
3129 	if (!err) {
3130 		printk(KERN_CRIT "NTFS: Aborting NTFS filesystem driver "
3131 				"registration...\n");
3132 		err = -ENOMEM;
3133 	}
3134 	return err;
3135 }
3136 
3137 static void __exit exit_ntfs_fs(void)
3138 {
3139 	int err = 0;
3140 
3141 	ntfs_debug("Unregistering NTFS driver.");
3142 
3143 	unregister_filesystem(&ntfs_fs_type);
3144 
3145 	if (kmem_cache_destroy(ntfs_big_inode_cache) && (err = 1))
3146 		printk(KERN_CRIT "NTFS: Failed to destory %s.\n",
3147 				ntfs_big_inode_cache_name);
3148 	if (kmem_cache_destroy(ntfs_inode_cache) && (err = 1))
3149 		printk(KERN_CRIT "NTFS: Failed to destory %s.\n",
3150 				ntfs_inode_cache_name);
3151 	if (kmem_cache_destroy(ntfs_name_cache) && (err = 1))
3152 		printk(KERN_CRIT "NTFS: Failed to destory %s.\n",
3153 				ntfs_name_cache_name);
3154 	if (kmem_cache_destroy(ntfs_attr_ctx_cache) && (err = 1))
3155 		printk(KERN_CRIT "NTFS: Failed to destory %s.\n",
3156 				ntfs_attr_ctx_cache_name);
3157 	if (kmem_cache_destroy(ntfs_index_ctx_cache) && (err = 1))
3158 		printk(KERN_CRIT "NTFS: Failed to destory %s.\n",
3159 				ntfs_index_ctx_cache_name);
3160 	if (err)
3161 		printk(KERN_CRIT "NTFS: This causes memory to leak! There is "
3162 				"probably a BUG in the driver! Please report "
3163 				"you saw this message to "
3164 				"linux-ntfs-dev@lists.sourceforge.net\n");
3165 	/* Unregister the ntfs sysctls. */
3166 	ntfs_sysctl(0);
3167 }
3168 
3169 MODULE_AUTHOR("Anton Altaparmakov <aia21@cantab.net>");
3170 MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2005 Anton Altaparmakov");
3171 MODULE_VERSION(NTFS_VERSION);
3172 MODULE_LICENSE("GPL");
3173 #ifdef DEBUG
3174 module_param(debug_msgs, bool, 0);
3175 MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3176 #endif
3177 
3178 module_init(init_ntfs_fs)
3179 module_exit(exit_ntfs_fs)
3180