xref: /openbmc/linux/fs/ntfs/super.c (revision d2999e1b)
1 /*
2  * super.c - NTFS kernel super block handling. Part of the Linux-NTFS project.
3  *
4  * Copyright (c) 2001-2012 Anton Altaparmakov and Tuxera Inc.
5  * Copyright (c) 2001,2002 Richard Russon
6  *
7  * This program/include file is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU General Public License as published
9  * by the Free Software Foundation; either version 2 of the License, or
10  * (at your option) any later version.
11  *
12  * This program/include file is distributed in the hope that it will be
13  * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
14  * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15  * GNU General Public License for more details.
16  *
17  * You should have received a copy of the GNU General Public License
18  * along with this program (in the main directory of the Linux-NTFS
19  * distribution in the file COPYING); if not, write to the Free Software
20  * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
21  */
22 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
23 
24 #include <linux/stddef.h>
25 #include <linux/init.h>
26 #include <linux/slab.h>
27 #include <linux/string.h>
28 #include <linux/spinlock.h>
29 #include <linux/blkdev.h>	/* For bdev_logical_block_size(). */
30 #include <linux/backing-dev.h>
31 #include <linux/buffer_head.h>
32 #include <linux/vfs.h>
33 #include <linux/moduleparam.h>
34 #include <linux/bitmap.h>
35 
36 #include "sysctl.h"
37 #include "logfile.h"
38 #include "quota.h"
39 #include "usnjrnl.h"
40 #include "dir.h"
41 #include "debug.h"
42 #include "index.h"
43 #include "inode.h"
44 #include "aops.h"
45 #include "layout.h"
46 #include "malloc.h"
47 #include "ntfs.h"
48 
49 /* Number of mounted filesystems which have compression enabled. */
50 static unsigned long ntfs_nr_compression_users;
51 
52 /* A global default upcase table and a corresponding reference count. */
53 static ntfschar *default_upcase;
54 static unsigned long ntfs_nr_upcase_users;
55 
56 /* Error constants/strings used in inode.c::ntfs_show_options(). */
57 typedef enum {
58 	/* One of these must be present, default is ON_ERRORS_CONTINUE. */
59 	ON_ERRORS_PANIC			= 0x01,
60 	ON_ERRORS_REMOUNT_RO		= 0x02,
61 	ON_ERRORS_CONTINUE		= 0x04,
62 	/* Optional, can be combined with any of the above. */
63 	ON_ERRORS_RECOVER		= 0x10,
64 } ON_ERRORS_ACTIONS;
65 
66 const option_t on_errors_arr[] = {
67 	{ ON_ERRORS_PANIC,	"panic" },
68 	{ ON_ERRORS_REMOUNT_RO,	"remount-ro", },
69 	{ ON_ERRORS_CONTINUE,	"continue", },
70 	{ ON_ERRORS_RECOVER,	"recover" },
71 	{ 0,			NULL }
72 };
73 
74 /**
75  * simple_getbool -
76  *
77  * Copied from old ntfs driver (which copied from vfat driver).
78  */
79 static int simple_getbool(char *s, bool *setval)
80 {
81 	if (s) {
82 		if (!strcmp(s, "1") || !strcmp(s, "yes") || !strcmp(s, "true"))
83 			*setval = true;
84 		else if (!strcmp(s, "0") || !strcmp(s, "no") ||
85 							!strcmp(s, "false"))
86 			*setval = false;
87 		else
88 			return 0;
89 	} else
90 		*setval = true;
91 	return 1;
92 }
93 
94 /**
95  * parse_options - parse the (re)mount options
96  * @vol:	ntfs volume
97  * @opt:	string containing the (re)mount options
98  *
99  * Parse the recognized options in @opt for the ntfs volume described by @vol.
100  */
101 static bool parse_options(ntfs_volume *vol, char *opt)
102 {
103 	char *p, *v, *ov;
104 	static char *utf8 = "utf8";
105 	int errors = 0, sloppy = 0;
106 	kuid_t uid = INVALID_UID;
107 	kgid_t gid = INVALID_GID;
108 	umode_t fmask = (umode_t)-1, dmask = (umode_t)-1;
109 	int mft_zone_multiplier = -1, on_errors = -1;
110 	int show_sys_files = -1, case_sensitive = -1, disable_sparse = -1;
111 	struct nls_table *nls_map = NULL, *old_nls;
112 
113 	/* I am lazy... (-8 */
114 #define NTFS_GETOPT_WITH_DEFAULT(option, variable, default_value)	\
115 	if (!strcmp(p, option)) {					\
116 		if (!v || !*v)						\
117 			variable = default_value;			\
118 		else {							\
119 			variable = simple_strtoul(ov = v, &v, 0);	\
120 			if (*v)						\
121 				goto needs_val;				\
122 		}							\
123 	}
124 #define NTFS_GETOPT(option, variable)					\
125 	if (!strcmp(p, option)) {					\
126 		if (!v || !*v)						\
127 			goto needs_arg;					\
128 		variable = simple_strtoul(ov = v, &v, 0);		\
129 		if (*v)							\
130 			goto needs_val;					\
131 	}
132 #define NTFS_GETOPT_UID(option, variable)				\
133 	if (!strcmp(p, option)) {					\
134 		uid_t uid_value;					\
135 		if (!v || !*v)						\
136 			goto needs_arg;					\
137 		uid_value = simple_strtoul(ov = v, &v, 0);		\
138 		if (*v)							\
139 			goto needs_val;					\
140 		variable = make_kuid(current_user_ns(), uid_value);	\
141 		if (!uid_valid(variable))				\
142 			goto needs_val;					\
143 	}
144 #define NTFS_GETOPT_GID(option, variable)				\
145 	if (!strcmp(p, option)) {					\
146 		gid_t gid_value;					\
147 		if (!v || !*v)						\
148 			goto needs_arg;					\
149 		gid_value = simple_strtoul(ov = v, &v, 0);		\
150 		if (*v)							\
151 			goto needs_val;					\
152 		variable = make_kgid(current_user_ns(), gid_value);	\
153 		if (!gid_valid(variable))				\
154 			goto needs_val;					\
155 	}
156 #define NTFS_GETOPT_OCTAL(option, variable)				\
157 	if (!strcmp(p, option)) {					\
158 		if (!v || !*v)						\
159 			goto needs_arg;					\
160 		variable = simple_strtoul(ov = v, &v, 8);		\
161 		if (*v)							\
162 			goto needs_val;					\
163 	}
164 #define NTFS_GETOPT_BOOL(option, variable)				\
165 	if (!strcmp(p, option)) {					\
166 		bool val;						\
167 		if (!simple_getbool(v, &val))				\
168 			goto needs_bool;				\
169 		variable = val;						\
170 	}
171 #define NTFS_GETOPT_OPTIONS_ARRAY(option, variable, opt_array)		\
172 	if (!strcmp(p, option)) {					\
173 		int _i;							\
174 		if (!v || !*v)						\
175 			goto needs_arg;					\
176 		ov = v;							\
177 		if (variable == -1)					\
178 			variable = 0;					\
179 		for (_i = 0; opt_array[_i].str && *opt_array[_i].str; _i++) \
180 			if (!strcmp(opt_array[_i].str, v)) {		\
181 				variable |= opt_array[_i].val;		\
182 				break;					\
183 			}						\
184 		if (!opt_array[_i].str || !*opt_array[_i].str)		\
185 			goto needs_val;					\
186 	}
187 	if (!opt || !*opt)
188 		goto no_mount_options;
189 	ntfs_debug("Entering with mount options string: %s", opt);
190 	while ((p = strsep(&opt, ","))) {
191 		if ((v = strchr(p, '=')))
192 			*v++ = 0;
193 		NTFS_GETOPT_UID("uid", uid)
194 		else NTFS_GETOPT_GID("gid", gid)
195 		else NTFS_GETOPT_OCTAL("umask", fmask = dmask)
196 		else NTFS_GETOPT_OCTAL("fmask", fmask)
197 		else NTFS_GETOPT_OCTAL("dmask", dmask)
198 		else NTFS_GETOPT("mft_zone_multiplier", mft_zone_multiplier)
199 		else NTFS_GETOPT_WITH_DEFAULT("sloppy", sloppy, true)
200 		else NTFS_GETOPT_BOOL("show_sys_files", show_sys_files)
201 		else NTFS_GETOPT_BOOL("case_sensitive", case_sensitive)
202 		else NTFS_GETOPT_BOOL("disable_sparse", disable_sparse)
203 		else NTFS_GETOPT_OPTIONS_ARRAY("errors", on_errors,
204 				on_errors_arr)
205 		else if (!strcmp(p, "posix") || !strcmp(p, "show_inodes"))
206 			ntfs_warning(vol->sb, "Ignoring obsolete option %s.",
207 					p);
208 		else if (!strcmp(p, "nls") || !strcmp(p, "iocharset")) {
209 			if (!strcmp(p, "iocharset"))
210 				ntfs_warning(vol->sb, "Option iocharset is "
211 						"deprecated. Please use "
212 						"option nls=<charsetname> in "
213 						"the future.");
214 			if (!v || !*v)
215 				goto needs_arg;
216 use_utf8:
217 			old_nls = nls_map;
218 			nls_map = load_nls(v);
219 			if (!nls_map) {
220 				if (!old_nls) {
221 					ntfs_error(vol->sb, "NLS character set "
222 							"%s not found.", v);
223 					return false;
224 				}
225 				ntfs_error(vol->sb, "NLS character set %s not "
226 						"found. Using previous one %s.",
227 						v, old_nls->charset);
228 				nls_map = old_nls;
229 			} else /* nls_map */ {
230 				unload_nls(old_nls);
231 			}
232 		} else if (!strcmp(p, "utf8")) {
233 			bool val = false;
234 			ntfs_warning(vol->sb, "Option utf8 is no longer "
235 				   "supported, using option nls=utf8. Please "
236 				   "use option nls=utf8 in the future and "
237 				   "make sure utf8 is compiled either as a "
238 				   "module or into the kernel.");
239 			if (!v || !*v)
240 				val = true;
241 			else if (!simple_getbool(v, &val))
242 				goto needs_bool;
243 			if (val) {
244 				v = utf8;
245 				goto use_utf8;
246 			}
247 		} else {
248 			ntfs_error(vol->sb, "Unrecognized mount option %s.", p);
249 			if (errors < INT_MAX)
250 				errors++;
251 		}
252 #undef NTFS_GETOPT_OPTIONS_ARRAY
253 #undef NTFS_GETOPT_BOOL
254 #undef NTFS_GETOPT
255 #undef NTFS_GETOPT_WITH_DEFAULT
256 	}
257 no_mount_options:
258 	if (errors && !sloppy)
259 		return false;
260 	if (sloppy)
261 		ntfs_warning(vol->sb, "Sloppy option given. Ignoring "
262 				"unrecognized mount option(s) and continuing.");
263 	/* Keep this first! */
264 	if (on_errors != -1) {
265 		if (!on_errors) {
266 			ntfs_error(vol->sb, "Invalid errors option argument "
267 					"or bug in options parser.");
268 			return false;
269 		}
270 	}
271 	if (nls_map) {
272 		if (vol->nls_map && vol->nls_map != nls_map) {
273 			ntfs_error(vol->sb, "Cannot change NLS character set "
274 					"on remount.");
275 			return false;
276 		} /* else (!vol->nls_map) */
277 		ntfs_debug("Using NLS character set %s.", nls_map->charset);
278 		vol->nls_map = nls_map;
279 	} else /* (!nls_map) */ {
280 		if (!vol->nls_map) {
281 			vol->nls_map = load_nls_default();
282 			if (!vol->nls_map) {
283 				ntfs_error(vol->sb, "Failed to load default "
284 						"NLS character set.");
285 				return false;
286 			}
287 			ntfs_debug("Using default NLS character set (%s).",
288 					vol->nls_map->charset);
289 		}
290 	}
291 	if (mft_zone_multiplier != -1) {
292 		if (vol->mft_zone_multiplier && vol->mft_zone_multiplier !=
293 				mft_zone_multiplier) {
294 			ntfs_error(vol->sb, "Cannot change mft_zone_multiplier "
295 					"on remount.");
296 			return false;
297 		}
298 		if (mft_zone_multiplier < 1 || mft_zone_multiplier > 4) {
299 			ntfs_error(vol->sb, "Invalid mft_zone_multiplier. "
300 					"Using default value, i.e. 1.");
301 			mft_zone_multiplier = 1;
302 		}
303 		vol->mft_zone_multiplier = mft_zone_multiplier;
304 	}
305 	if (!vol->mft_zone_multiplier)
306 		vol->mft_zone_multiplier = 1;
307 	if (on_errors != -1)
308 		vol->on_errors = on_errors;
309 	if (!vol->on_errors || vol->on_errors == ON_ERRORS_RECOVER)
310 		vol->on_errors |= ON_ERRORS_CONTINUE;
311 	if (uid_valid(uid))
312 		vol->uid = uid;
313 	if (gid_valid(gid))
314 		vol->gid = gid;
315 	if (fmask != (umode_t)-1)
316 		vol->fmask = fmask;
317 	if (dmask != (umode_t)-1)
318 		vol->dmask = dmask;
319 	if (show_sys_files != -1) {
320 		if (show_sys_files)
321 			NVolSetShowSystemFiles(vol);
322 		else
323 			NVolClearShowSystemFiles(vol);
324 	}
325 	if (case_sensitive != -1) {
326 		if (case_sensitive)
327 			NVolSetCaseSensitive(vol);
328 		else
329 			NVolClearCaseSensitive(vol);
330 	}
331 	if (disable_sparse != -1) {
332 		if (disable_sparse)
333 			NVolClearSparseEnabled(vol);
334 		else {
335 			if (!NVolSparseEnabled(vol) &&
336 					vol->major_ver && vol->major_ver < 3)
337 				ntfs_warning(vol->sb, "Not enabling sparse "
338 						"support due to NTFS volume "
339 						"version %i.%i (need at least "
340 						"version 3.0).", vol->major_ver,
341 						vol->minor_ver);
342 			else
343 				NVolSetSparseEnabled(vol);
344 		}
345 	}
346 	return true;
347 needs_arg:
348 	ntfs_error(vol->sb, "The %s option requires an argument.", p);
349 	return false;
350 needs_bool:
351 	ntfs_error(vol->sb, "The %s option requires a boolean argument.", p);
352 	return false;
353 needs_val:
354 	ntfs_error(vol->sb, "Invalid %s option argument: %s", p, ov);
355 	return false;
356 }
357 
358 #ifdef NTFS_RW
359 
360 /**
361  * ntfs_write_volume_flags - write new flags to the volume information flags
362  * @vol:	ntfs volume on which to modify the flags
363  * @flags:	new flags value for the volume information flags
364  *
365  * Internal function.  You probably want to use ntfs_{set,clear}_volume_flags()
366  * instead (see below).
367  *
368  * Replace the volume information flags on the volume @vol with the value
369  * supplied in @flags.  Note, this overwrites the volume information flags, so
370  * make sure to combine the flags you want to modify with the old flags and use
371  * the result when calling ntfs_write_volume_flags().
372  *
373  * Return 0 on success and -errno on error.
374  */
375 static int ntfs_write_volume_flags(ntfs_volume *vol, const VOLUME_FLAGS flags)
376 {
377 	ntfs_inode *ni = NTFS_I(vol->vol_ino);
378 	MFT_RECORD *m;
379 	VOLUME_INFORMATION *vi;
380 	ntfs_attr_search_ctx *ctx;
381 	int err;
382 
383 	ntfs_debug("Entering, old flags = 0x%x, new flags = 0x%x.",
384 			le16_to_cpu(vol->vol_flags), le16_to_cpu(flags));
385 	if (vol->vol_flags == flags)
386 		goto done;
387 	BUG_ON(!ni);
388 	m = map_mft_record(ni);
389 	if (IS_ERR(m)) {
390 		err = PTR_ERR(m);
391 		goto err_out;
392 	}
393 	ctx = ntfs_attr_get_search_ctx(ni, m);
394 	if (!ctx) {
395 		err = -ENOMEM;
396 		goto put_unm_err_out;
397 	}
398 	err = ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
399 			ctx);
400 	if (err)
401 		goto put_unm_err_out;
402 	vi = (VOLUME_INFORMATION*)((u8*)ctx->attr +
403 			le16_to_cpu(ctx->attr->data.resident.value_offset));
404 	vol->vol_flags = vi->flags = flags;
405 	flush_dcache_mft_record_page(ctx->ntfs_ino);
406 	mark_mft_record_dirty(ctx->ntfs_ino);
407 	ntfs_attr_put_search_ctx(ctx);
408 	unmap_mft_record(ni);
409 done:
410 	ntfs_debug("Done.");
411 	return 0;
412 put_unm_err_out:
413 	if (ctx)
414 		ntfs_attr_put_search_ctx(ctx);
415 	unmap_mft_record(ni);
416 err_out:
417 	ntfs_error(vol->sb, "Failed with error code %i.", -err);
418 	return err;
419 }
420 
421 /**
422  * ntfs_set_volume_flags - set bits in the volume information flags
423  * @vol:	ntfs volume on which to modify the flags
424  * @flags:	flags to set on the volume
425  *
426  * Set the bits in @flags in the volume information flags on the volume @vol.
427  *
428  * Return 0 on success and -errno on error.
429  */
430 static inline int ntfs_set_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
431 {
432 	flags &= VOLUME_FLAGS_MASK;
433 	return ntfs_write_volume_flags(vol, vol->vol_flags | flags);
434 }
435 
436 /**
437  * ntfs_clear_volume_flags - clear bits in the volume information flags
438  * @vol:	ntfs volume on which to modify the flags
439  * @flags:	flags to clear on the volume
440  *
441  * Clear the bits in @flags in the volume information flags on the volume @vol.
442  *
443  * Return 0 on success and -errno on error.
444  */
445 static inline int ntfs_clear_volume_flags(ntfs_volume *vol, VOLUME_FLAGS flags)
446 {
447 	flags &= VOLUME_FLAGS_MASK;
448 	flags = vol->vol_flags & cpu_to_le16(~le16_to_cpu(flags));
449 	return ntfs_write_volume_flags(vol, flags);
450 }
451 
452 #endif /* NTFS_RW */
453 
454 /**
455  * ntfs_remount - change the mount options of a mounted ntfs filesystem
456  * @sb:		superblock of mounted ntfs filesystem
457  * @flags:	remount flags
458  * @opt:	remount options string
459  *
460  * Change the mount options of an already mounted ntfs filesystem.
461  *
462  * NOTE:  The VFS sets the @sb->s_flags remount flags to @flags after
463  * ntfs_remount() returns successfully (i.e. returns 0).  Otherwise,
464  * @sb->s_flags are not changed.
465  */
466 static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
467 {
468 	ntfs_volume *vol = NTFS_SB(sb);
469 
470 	ntfs_debug("Entering with remount options string: %s", opt);
471 
472 	sync_filesystem(sb);
473 
474 #ifndef NTFS_RW
475 	/* For read-only compiled driver, enforce read-only flag. */
476 	*flags |= MS_RDONLY;
477 #else /* NTFS_RW */
478 	/*
479 	 * For the read-write compiled driver, if we are remounting read-write,
480 	 * make sure there are no volume errors and that no unsupported volume
481 	 * flags are set.  Also, empty the logfile journal as it would become
482 	 * stale as soon as something is written to the volume and mark the
483 	 * volume dirty so that chkdsk is run if the volume is not umounted
484 	 * cleanly.  Finally, mark the quotas out of date so Windows rescans
485 	 * the volume on boot and updates them.
486 	 *
487 	 * When remounting read-only, mark the volume clean if no volume errors
488 	 * have occurred.
489 	 */
490 	if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
491 		static const char *es = ".  Cannot remount read-write.";
492 
493 		/* Remounting read-write. */
494 		if (NVolErrors(vol)) {
495 			ntfs_error(sb, "Volume has errors and is read-only%s",
496 					es);
497 			return -EROFS;
498 		}
499 		if (vol->vol_flags & VOLUME_IS_DIRTY) {
500 			ntfs_error(sb, "Volume is dirty and read-only%s", es);
501 			return -EROFS;
502 		}
503 		if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
504 			ntfs_error(sb, "Volume has been modified by chkdsk "
505 					"and is read-only%s", es);
506 			return -EROFS;
507 		}
508 		if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
509 			ntfs_error(sb, "Volume has unsupported flags set "
510 					"(0x%x) and is read-only%s",
511 					(unsigned)le16_to_cpu(vol->vol_flags),
512 					es);
513 			return -EROFS;
514 		}
515 		if (ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
516 			ntfs_error(sb, "Failed to set dirty bit in volume "
517 					"information flags%s", es);
518 			return -EROFS;
519 		}
520 #if 0
521 		// TODO: Enable this code once we start modifying anything that
522 		//	 is different between NTFS 1.2 and 3.x...
523 		/* Set NT4 compatibility flag on newer NTFS version volumes. */
524 		if ((vol->major_ver > 1)) {
525 			if (ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
526 				ntfs_error(sb, "Failed to set NT4 "
527 						"compatibility flag%s", es);
528 				NVolSetErrors(vol);
529 				return -EROFS;
530 			}
531 		}
532 #endif
533 		if (!ntfs_empty_logfile(vol->logfile_ino)) {
534 			ntfs_error(sb, "Failed to empty journal $LogFile%s",
535 					es);
536 			NVolSetErrors(vol);
537 			return -EROFS;
538 		}
539 		if (!ntfs_mark_quotas_out_of_date(vol)) {
540 			ntfs_error(sb, "Failed to mark quotas out of date%s",
541 					es);
542 			NVolSetErrors(vol);
543 			return -EROFS;
544 		}
545 		if (!ntfs_stamp_usnjrnl(vol)) {
546 			ntfs_error(sb, "Failed to stamp transation log "
547 					"($UsnJrnl)%s", es);
548 			NVolSetErrors(vol);
549 			return -EROFS;
550 		}
551 	} else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) {
552 		/* Remounting read-only. */
553 		if (!NVolErrors(vol)) {
554 			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
555 				ntfs_warning(sb, "Failed to clear dirty bit "
556 						"in volume information "
557 						"flags.  Run chkdsk.");
558 		}
559 	}
560 #endif /* NTFS_RW */
561 
562 	// TODO: Deal with *flags.
563 
564 	if (!parse_options(vol, opt))
565 		return -EINVAL;
566 
567 	ntfs_debug("Done.");
568 	return 0;
569 }
570 
571 /**
572  * is_boot_sector_ntfs - check whether a boot sector is a valid NTFS boot sector
573  * @sb:		Super block of the device to which @b belongs.
574  * @b:		Boot sector of device @sb to check.
575  * @silent:	If 'true', all output will be silenced.
576  *
577  * is_boot_sector_ntfs() checks whether the boot sector @b is a valid NTFS boot
578  * sector. Returns 'true' if it is valid and 'false' if not.
579  *
580  * @sb is only needed for warning/error output, i.e. it can be NULL when silent
581  * is 'true'.
582  */
583 static bool is_boot_sector_ntfs(const struct super_block *sb,
584 		const NTFS_BOOT_SECTOR *b, const bool silent)
585 {
586 	/*
587 	 * Check that checksum == sum of u32 values from b to the checksum
588 	 * field.  If checksum is zero, no checking is done.  We will work when
589 	 * the checksum test fails, since some utilities update the boot sector
590 	 * ignoring the checksum which leaves the checksum out-of-date.  We
591 	 * report a warning if this is the case.
592 	 */
593 	if ((void*)b < (void*)&b->checksum && b->checksum && !silent) {
594 		le32 *u;
595 		u32 i;
596 
597 		for (i = 0, u = (le32*)b; u < (le32*)(&b->checksum); ++u)
598 			i += le32_to_cpup(u);
599 		if (le32_to_cpu(b->checksum) != i)
600 			ntfs_warning(sb, "Invalid boot sector checksum.");
601 	}
602 	/* Check OEMidentifier is "NTFS    " */
603 	if (b->oem_id != magicNTFS)
604 		goto not_ntfs;
605 	/* Check bytes per sector value is between 256 and 4096. */
606 	if (le16_to_cpu(b->bpb.bytes_per_sector) < 0x100 ||
607 			le16_to_cpu(b->bpb.bytes_per_sector) > 0x1000)
608 		goto not_ntfs;
609 	/* Check sectors per cluster value is valid. */
610 	switch (b->bpb.sectors_per_cluster) {
611 	case 1: case 2: case 4: case 8: case 16: case 32: case 64: case 128:
612 		break;
613 	default:
614 		goto not_ntfs;
615 	}
616 	/* Check the cluster size is not above the maximum (64kiB). */
617 	if ((u32)le16_to_cpu(b->bpb.bytes_per_sector) *
618 			b->bpb.sectors_per_cluster > NTFS_MAX_CLUSTER_SIZE)
619 		goto not_ntfs;
620 	/* Check reserved/unused fields are really zero. */
621 	if (le16_to_cpu(b->bpb.reserved_sectors) ||
622 			le16_to_cpu(b->bpb.root_entries) ||
623 			le16_to_cpu(b->bpb.sectors) ||
624 			le16_to_cpu(b->bpb.sectors_per_fat) ||
625 			le32_to_cpu(b->bpb.large_sectors) || b->bpb.fats)
626 		goto not_ntfs;
627 	/* Check clusters per file mft record value is valid. */
628 	if ((u8)b->clusters_per_mft_record < 0xe1 ||
629 			(u8)b->clusters_per_mft_record > 0xf7)
630 		switch (b->clusters_per_mft_record) {
631 		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
632 			break;
633 		default:
634 			goto not_ntfs;
635 		}
636 	/* Check clusters per index block value is valid. */
637 	if ((u8)b->clusters_per_index_record < 0xe1 ||
638 			(u8)b->clusters_per_index_record > 0xf7)
639 		switch (b->clusters_per_index_record) {
640 		case 1: case 2: case 4: case 8: case 16: case 32: case 64:
641 			break;
642 		default:
643 			goto not_ntfs;
644 		}
645 	/*
646 	 * Check for valid end of sector marker. We will work without it, but
647 	 * many BIOSes will refuse to boot from a bootsector if the magic is
648 	 * incorrect, so we emit a warning.
649 	 */
650 	if (!silent && b->end_of_sector_marker != cpu_to_le16(0xaa55))
651 		ntfs_warning(sb, "Invalid end of sector marker.");
652 	return true;
653 not_ntfs:
654 	return false;
655 }
656 
657 /**
658  * read_ntfs_boot_sector - read the NTFS boot sector of a device
659  * @sb:		super block of device to read the boot sector from
660  * @silent:	if true, suppress all output
661  *
662  * Reads the boot sector from the device and validates it. If that fails, tries
663  * to read the backup boot sector, first from the end of the device a-la NT4 and
664  * later and then from the middle of the device a-la NT3.51 and before.
665  *
666  * If a valid boot sector is found but it is not the primary boot sector, we
667  * repair the primary boot sector silently (unless the device is read-only or
668  * the primary boot sector is not accessible).
669  *
670  * NOTE: To call this function, @sb must have the fields s_dev, the ntfs super
671  * block (u.ntfs_sb), nr_blocks and the device flags (s_flags) initialized
672  * to their respective values.
673  *
674  * Return the unlocked buffer head containing the boot sector or NULL on error.
675  */
676 static struct buffer_head *read_ntfs_boot_sector(struct super_block *sb,
677 		const int silent)
678 {
679 	const char *read_err_str = "Unable to read %s boot sector.";
680 	struct buffer_head *bh_primary, *bh_backup;
681 	sector_t nr_blocks = NTFS_SB(sb)->nr_blocks;
682 
683 	/* Try to read primary boot sector. */
684 	if ((bh_primary = sb_bread(sb, 0))) {
685 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
686 				bh_primary->b_data, silent))
687 			return bh_primary;
688 		if (!silent)
689 			ntfs_error(sb, "Primary boot sector is invalid.");
690 	} else if (!silent)
691 		ntfs_error(sb, read_err_str, "primary");
692 	if (!(NTFS_SB(sb)->on_errors & ON_ERRORS_RECOVER)) {
693 		if (bh_primary)
694 			brelse(bh_primary);
695 		if (!silent)
696 			ntfs_error(sb, "Mount option errors=recover not used. "
697 					"Aborting without trying to recover.");
698 		return NULL;
699 	}
700 	/* Try to read NT4+ backup boot sector. */
701 	if ((bh_backup = sb_bread(sb, nr_blocks - 1))) {
702 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
703 				bh_backup->b_data, silent))
704 			goto hotfix_primary_boot_sector;
705 		brelse(bh_backup);
706 	} else if (!silent)
707 		ntfs_error(sb, read_err_str, "backup");
708 	/* Try to read NT3.51- backup boot sector. */
709 	if ((bh_backup = sb_bread(sb, nr_blocks >> 1))) {
710 		if (is_boot_sector_ntfs(sb, (NTFS_BOOT_SECTOR*)
711 				bh_backup->b_data, silent))
712 			goto hotfix_primary_boot_sector;
713 		if (!silent)
714 			ntfs_error(sb, "Could not find a valid backup boot "
715 					"sector.");
716 		brelse(bh_backup);
717 	} else if (!silent)
718 		ntfs_error(sb, read_err_str, "backup");
719 	/* We failed. Cleanup and return. */
720 	if (bh_primary)
721 		brelse(bh_primary);
722 	return NULL;
723 hotfix_primary_boot_sector:
724 	if (bh_primary) {
725 		/*
726 		 * If we managed to read sector zero and the volume is not
727 		 * read-only, copy the found, valid backup boot sector to the
728 		 * primary boot sector.  Note we only copy the actual boot
729 		 * sector structure, not the actual whole device sector as that
730 		 * may be bigger and would potentially damage the $Boot system
731 		 * file (FIXME: Would be nice to know if the backup boot sector
732 		 * on a large sector device contains the whole boot loader or
733 		 * just the first 512 bytes).
734 		 */
735 		if (!(sb->s_flags & MS_RDONLY)) {
736 			ntfs_warning(sb, "Hot-fix: Recovering invalid primary "
737 					"boot sector from backup copy.");
738 			memcpy(bh_primary->b_data, bh_backup->b_data,
739 					NTFS_BLOCK_SIZE);
740 			mark_buffer_dirty(bh_primary);
741 			sync_dirty_buffer(bh_primary);
742 			if (buffer_uptodate(bh_primary)) {
743 				brelse(bh_backup);
744 				return bh_primary;
745 			}
746 			ntfs_error(sb, "Hot-fix: Device write error while "
747 					"recovering primary boot sector.");
748 		} else {
749 			ntfs_warning(sb, "Hot-fix: Recovery of primary boot "
750 					"sector failed: Read-only mount.");
751 		}
752 		brelse(bh_primary);
753 	}
754 	ntfs_warning(sb, "Using backup boot sector.");
755 	return bh_backup;
756 }
757 
758 /**
759  * parse_ntfs_boot_sector - parse the boot sector and store the data in @vol
760  * @vol:	volume structure to initialise with data from boot sector
761  * @b:		boot sector to parse
762  *
763  * Parse the ntfs boot sector @b and store all imporant information therein in
764  * the ntfs super block @vol.  Return 'true' on success and 'false' on error.
765  */
766 static bool parse_ntfs_boot_sector(ntfs_volume *vol, const NTFS_BOOT_SECTOR *b)
767 {
768 	unsigned int sectors_per_cluster_bits, nr_hidden_sects;
769 	int clusters_per_mft_record, clusters_per_index_record;
770 	s64 ll;
771 
772 	vol->sector_size = le16_to_cpu(b->bpb.bytes_per_sector);
773 	vol->sector_size_bits = ffs(vol->sector_size) - 1;
774 	ntfs_debug("vol->sector_size = %i (0x%x)", vol->sector_size,
775 			vol->sector_size);
776 	ntfs_debug("vol->sector_size_bits = %i (0x%x)", vol->sector_size_bits,
777 			vol->sector_size_bits);
778 	if (vol->sector_size < vol->sb->s_blocksize) {
779 		ntfs_error(vol->sb, "Sector size (%i) is smaller than the "
780 				"device block size (%lu).  This is not "
781 				"supported.  Sorry.", vol->sector_size,
782 				vol->sb->s_blocksize);
783 		return false;
784 	}
785 	ntfs_debug("sectors_per_cluster = 0x%x", b->bpb.sectors_per_cluster);
786 	sectors_per_cluster_bits = ffs(b->bpb.sectors_per_cluster) - 1;
787 	ntfs_debug("sectors_per_cluster_bits = 0x%x",
788 			sectors_per_cluster_bits);
789 	nr_hidden_sects = le32_to_cpu(b->bpb.hidden_sectors);
790 	ntfs_debug("number of hidden sectors = 0x%x", nr_hidden_sects);
791 	vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;
792 	vol->cluster_size_mask = vol->cluster_size - 1;
793 	vol->cluster_size_bits = ffs(vol->cluster_size) - 1;
794 	ntfs_debug("vol->cluster_size = %i (0x%x)", vol->cluster_size,
795 			vol->cluster_size);
796 	ntfs_debug("vol->cluster_size_mask = 0x%x", vol->cluster_size_mask);
797 	ntfs_debug("vol->cluster_size_bits = %i", vol->cluster_size_bits);
798 	if (vol->cluster_size < vol->sector_size) {
799 		ntfs_error(vol->sb, "Cluster size (%i) is smaller than the "
800 				"sector size (%i).  This is not supported.  "
801 				"Sorry.", vol->cluster_size, vol->sector_size);
802 		return false;
803 	}
804 	clusters_per_mft_record = b->clusters_per_mft_record;
805 	ntfs_debug("clusters_per_mft_record = %i (0x%x)",
806 			clusters_per_mft_record, clusters_per_mft_record);
807 	if (clusters_per_mft_record > 0)
808 		vol->mft_record_size = vol->cluster_size <<
809 				(ffs(clusters_per_mft_record) - 1);
810 	else
811 		/*
812 		 * When mft_record_size < cluster_size, clusters_per_mft_record
813 		 * = -log2(mft_record_size) bytes. mft_record_size normaly is
814 		 * 1024 bytes, which is encoded as 0xF6 (-10 in decimal).
815 		 */
816 		vol->mft_record_size = 1 << -clusters_per_mft_record;
817 	vol->mft_record_size_mask = vol->mft_record_size - 1;
818 	vol->mft_record_size_bits = ffs(vol->mft_record_size) - 1;
819 	ntfs_debug("vol->mft_record_size = %i (0x%x)", vol->mft_record_size,
820 			vol->mft_record_size);
821 	ntfs_debug("vol->mft_record_size_mask = 0x%x",
822 			vol->mft_record_size_mask);
823 	ntfs_debug("vol->mft_record_size_bits = %i (0x%x)",
824 			vol->mft_record_size_bits, vol->mft_record_size_bits);
825 	/*
826 	 * We cannot support mft record sizes above the PAGE_CACHE_SIZE since
827 	 * we store $MFT/$DATA, the table of mft records in the page cache.
828 	 */
829 	if (vol->mft_record_size > PAGE_CACHE_SIZE) {
830 		ntfs_error(vol->sb, "Mft record size (%i) exceeds the "
831 				"PAGE_CACHE_SIZE on your system (%lu).  "
832 				"This is not supported.  Sorry.",
833 				vol->mft_record_size, PAGE_CACHE_SIZE);
834 		return false;
835 	}
836 	/* We cannot support mft record sizes below the sector size. */
837 	if (vol->mft_record_size < vol->sector_size) {
838 		ntfs_error(vol->sb, "Mft record size (%i) is smaller than the "
839 				"sector size (%i).  This is not supported.  "
840 				"Sorry.", vol->mft_record_size,
841 				vol->sector_size);
842 		return false;
843 	}
844 	clusters_per_index_record = b->clusters_per_index_record;
845 	ntfs_debug("clusters_per_index_record = %i (0x%x)",
846 			clusters_per_index_record, clusters_per_index_record);
847 	if (clusters_per_index_record > 0)
848 		vol->index_record_size = vol->cluster_size <<
849 				(ffs(clusters_per_index_record) - 1);
850 	else
851 		/*
852 		 * When index_record_size < cluster_size,
853 		 * clusters_per_index_record = -log2(index_record_size) bytes.
854 		 * index_record_size normaly equals 4096 bytes, which is
855 		 * encoded as 0xF4 (-12 in decimal).
856 		 */
857 		vol->index_record_size = 1 << -clusters_per_index_record;
858 	vol->index_record_size_mask = vol->index_record_size - 1;
859 	vol->index_record_size_bits = ffs(vol->index_record_size) - 1;
860 	ntfs_debug("vol->index_record_size = %i (0x%x)",
861 			vol->index_record_size, vol->index_record_size);
862 	ntfs_debug("vol->index_record_size_mask = 0x%x",
863 			vol->index_record_size_mask);
864 	ntfs_debug("vol->index_record_size_bits = %i (0x%x)",
865 			vol->index_record_size_bits,
866 			vol->index_record_size_bits);
867 	/* We cannot support index record sizes below the sector size. */
868 	if (vol->index_record_size < vol->sector_size) {
869 		ntfs_error(vol->sb, "Index record size (%i) is smaller than "
870 				"the sector size (%i).  This is not "
871 				"supported.  Sorry.", vol->index_record_size,
872 				vol->sector_size);
873 		return false;
874 	}
875 	/*
876 	 * Get the size of the volume in clusters and check for 64-bit-ness.
877 	 * Windows currently only uses 32 bits to save the clusters so we do
878 	 * the same as it is much faster on 32-bit CPUs.
879 	 */
880 	ll = sle64_to_cpu(b->number_of_sectors) >> sectors_per_cluster_bits;
881 	if ((u64)ll >= 1ULL << 32) {
882 		ntfs_error(vol->sb, "Cannot handle 64-bit clusters.  Sorry.");
883 		return false;
884 	}
885 	vol->nr_clusters = ll;
886 	ntfs_debug("vol->nr_clusters = 0x%llx", (long long)vol->nr_clusters);
887 	/*
888 	 * On an architecture where unsigned long is 32-bits, we restrict the
889 	 * volume size to 2TiB (2^41). On a 64-bit architecture, the compiler
890 	 * will hopefully optimize the whole check away.
891 	 */
892 	if (sizeof(unsigned long) < 8) {
893 		if ((ll << vol->cluster_size_bits) >= (1ULL << 41)) {
894 			ntfs_error(vol->sb, "Volume size (%lluTiB) is too "
895 					"large for this architecture.  "
896 					"Maximum supported is 2TiB.  Sorry.",
897 					(unsigned long long)ll >> (40 -
898 					vol->cluster_size_bits));
899 			return false;
900 		}
901 	}
902 	ll = sle64_to_cpu(b->mft_lcn);
903 	if (ll >= vol->nr_clusters) {
904 		ntfs_error(vol->sb, "MFT LCN (%lli, 0x%llx) is beyond end of "
905 				"volume.  Weird.", (unsigned long long)ll,
906 				(unsigned long long)ll);
907 		return false;
908 	}
909 	vol->mft_lcn = ll;
910 	ntfs_debug("vol->mft_lcn = 0x%llx", (long long)vol->mft_lcn);
911 	ll = sle64_to_cpu(b->mftmirr_lcn);
912 	if (ll >= vol->nr_clusters) {
913 		ntfs_error(vol->sb, "MFTMirr LCN (%lli, 0x%llx) is beyond end "
914 				"of volume.  Weird.", (unsigned long long)ll,
915 				(unsigned long long)ll);
916 		return false;
917 	}
918 	vol->mftmirr_lcn = ll;
919 	ntfs_debug("vol->mftmirr_lcn = 0x%llx", (long long)vol->mftmirr_lcn);
920 #ifdef NTFS_RW
921 	/*
922 	 * Work out the size of the mft mirror in number of mft records. If the
923 	 * cluster size is less than or equal to the size taken by four mft
924 	 * records, the mft mirror stores the first four mft records. If the
925 	 * cluster size is bigger than the size taken by four mft records, the
926 	 * mft mirror contains as many mft records as will fit into one
927 	 * cluster.
928 	 */
929 	if (vol->cluster_size <= (4 << vol->mft_record_size_bits))
930 		vol->mftmirr_size = 4;
931 	else
932 		vol->mftmirr_size = vol->cluster_size >>
933 				vol->mft_record_size_bits;
934 	ntfs_debug("vol->mftmirr_size = %i", vol->mftmirr_size);
935 #endif /* NTFS_RW */
936 	vol->serial_no = le64_to_cpu(b->volume_serial_number);
937 	ntfs_debug("vol->serial_no = 0x%llx",
938 			(unsigned long long)vol->serial_no);
939 	return true;
940 }
941 
942 /**
943  * ntfs_setup_allocators - initialize the cluster and mft allocators
944  * @vol:	volume structure for which to setup the allocators
945  *
946  * Setup the cluster (lcn) and mft allocators to the starting values.
947  */
948 static void ntfs_setup_allocators(ntfs_volume *vol)
949 {
950 #ifdef NTFS_RW
951 	LCN mft_zone_size, mft_lcn;
952 #endif /* NTFS_RW */
953 
954 	ntfs_debug("vol->mft_zone_multiplier = 0x%x",
955 			vol->mft_zone_multiplier);
956 #ifdef NTFS_RW
957 	/* Determine the size of the MFT zone. */
958 	mft_zone_size = vol->nr_clusters;
959 	switch (vol->mft_zone_multiplier) {  /* % of volume size in clusters */
960 	case 4:
961 		mft_zone_size >>= 1;			/* 50%   */
962 		break;
963 	case 3:
964 		mft_zone_size = (mft_zone_size +
965 				(mft_zone_size >> 1)) >> 2;	/* 37.5% */
966 		break;
967 	case 2:
968 		mft_zone_size >>= 2;			/* 25%   */
969 		break;
970 	/* case 1: */
971 	default:
972 		mft_zone_size >>= 3;			/* 12.5% */
973 		break;
974 	}
975 	/* Setup the mft zone. */
976 	vol->mft_zone_start = vol->mft_zone_pos = vol->mft_lcn;
977 	ntfs_debug("vol->mft_zone_pos = 0x%llx",
978 			(unsigned long long)vol->mft_zone_pos);
979 	/*
980 	 * Calculate the mft_lcn for an unmodified NTFS volume (see mkntfs
981 	 * source) and if the actual mft_lcn is in the expected place or even
982 	 * further to the front of the volume, extend the mft_zone to cover the
983 	 * beginning of the volume as well.  This is in order to protect the
984 	 * area reserved for the mft bitmap as well within the mft_zone itself.
985 	 * On non-standard volumes we do not protect it as the overhead would
986 	 * be higher than the speed increase we would get by doing it.
987 	 */
988 	mft_lcn = (8192 + 2 * vol->cluster_size - 1) / vol->cluster_size;
989 	if (mft_lcn * vol->cluster_size < 16 * 1024)
990 		mft_lcn = (16 * 1024 + vol->cluster_size - 1) /
991 				vol->cluster_size;
992 	if (vol->mft_zone_start <= mft_lcn)
993 		vol->mft_zone_start = 0;
994 	ntfs_debug("vol->mft_zone_start = 0x%llx",
995 			(unsigned long long)vol->mft_zone_start);
996 	/*
997 	 * Need to cap the mft zone on non-standard volumes so that it does
998 	 * not point outside the boundaries of the volume.  We do this by
999 	 * halving the zone size until we are inside the volume.
1000 	 */
1001 	vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
1002 	while (vol->mft_zone_end >= vol->nr_clusters) {
1003 		mft_zone_size >>= 1;
1004 		vol->mft_zone_end = vol->mft_lcn + mft_zone_size;
1005 	}
1006 	ntfs_debug("vol->mft_zone_end = 0x%llx",
1007 			(unsigned long long)vol->mft_zone_end);
1008 	/*
1009 	 * Set the current position within each data zone to the start of the
1010 	 * respective zone.
1011 	 */
1012 	vol->data1_zone_pos = vol->mft_zone_end;
1013 	ntfs_debug("vol->data1_zone_pos = 0x%llx",
1014 			(unsigned long long)vol->data1_zone_pos);
1015 	vol->data2_zone_pos = 0;
1016 	ntfs_debug("vol->data2_zone_pos = 0x%llx",
1017 			(unsigned long long)vol->data2_zone_pos);
1018 
1019 	/* Set the mft data allocation position to mft record 24. */
1020 	vol->mft_data_pos = 24;
1021 	ntfs_debug("vol->mft_data_pos = 0x%llx",
1022 			(unsigned long long)vol->mft_data_pos);
1023 #endif /* NTFS_RW */
1024 }
1025 
1026 #ifdef NTFS_RW
1027 
1028 /**
1029  * load_and_init_mft_mirror - load and setup the mft mirror inode for a volume
1030  * @vol:	ntfs super block describing device whose mft mirror to load
1031  *
1032  * Return 'true' on success or 'false' on error.
1033  */
1034 static bool load_and_init_mft_mirror(ntfs_volume *vol)
1035 {
1036 	struct inode *tmp_ino;
1037 	ntfs_inode *tmp_ni;
1038 
1039 	ntfs_debug("Entering.");
1040 	/* Get mft mirror inode. */
1041 	tmp_ino = ntfs_iget(vol->sb, FILE_MFTMirr);
1042 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1043 		if (!IS_ERR(tmp_ino))
1044 			iput(tmp_ino);
1045 		/* Caller will display error message. */
1046 		return false;
1047 	}
1048 	/*
1049 	 * Re-initialize some specifics about $MFTMirr's inode as
1050 	 * ntfs_read_inode() will have set up the default ones.
1051 	 */
1052 	/* Set uid and gid to root. */
1053 	tmp_ino->i_uid = GLOBAL_ROOT_UID;
1054 	tmp_ino->i_gid = GLOBAL_ROOT_GID;
1055 	/* Regular file.  No access for anyone. */
1056 	tmp_ino->i_mode = S_IFREG;
1057 	/* No VFS initiated operations allowed for $MFTMirr. */
1058 	tmp_ino->i_op = &ntfs_empty_inode_ops;
1059 	tmp_ino->i_fop = &ntfs_empty_file_ops;
1060 	/* Put in our special address space operations. */
1061 	tmp_ino->i_mapping->a_ops = &ntfs_mst_aops;
1062 	tmp_ni = NTFS_I(tmp_ino);
1063 	/* The $MFTMirr, like the $MFT is multi sector transfer protected. */
1064 	NInoSetMstProtected(tmp_ni);
1065 	NInoSetSparseDisabled(tmp_ni);
1066 	/*
1067 	 * Set up our little cheat allowing us to reuse the async read io
1068 	 * completion handler for directories.
1069 	 */
1070 	tmp_ni->itype.index.block_size = vol->mft_record_size;
1071 	tmp_ni->itype.index.block_size_bits = vol->mft_record_size_bits;
1072 	vol->mftmirr_ino = tmp_ino;
1073 	ntfs_debug("Done.");
1074 	return true;
1075 }
1076 
1077 /**
1078  * check_mft_mirror - compare contents of the mft mirror with the mft
1079  * @vol:	ntfs super block describing device whose mft mirror to check
1080  *
1081  * Return 'true' on success or 'false' on error.
1082  *
1083  * Note, this function also results in the mft mirror runlist being completely
1084  * mapped into memory.  The mft mirror write code requires this and will BUG()
1085  * should it find an unmapped runlist element.
1086  */
1087 static bool check_mft_mirror(ntfs_volume *vol)
1088 {
1089 	struct super_block *sb = vol->sb;
1090 	ntfs_inode *mirr_ni;
1091 	struct page *mft_page, *mirr_page;
1092 	u8 *kmft, *kmirr;
1093 	runlist_element *rl, rl2[2];
1094 	pgoff_t index;
1095 	int mrecs_per_page, i;
1096 
1097 	ntfs_debug("Entering.");
1098 	/* Compare contents of $MFT and $MFTMirr. */
1099 	mrecs_per_page = PAGE_CACHE_SIZE / vol->mft_record_size;
1100 	BUG_ON(!mrecs_per_page);
1101 	BUG_ON(!vol->mftmirr_size);
1102 	mft_page = mirr_page = NULL;
1103 	kmft = kmirr = NULL;
1104 	index = i = 0;
1105 	do {
1106 		u32 bytes;
1107 
1108 		/* Switch pages if necessary. */
1109 		if (!(i % mrecs_per_page)) {
1110 			if (index) {
1111 				ntfs_unmap_page(mft_page);
1112 				ntfs_unmap_page(mirr_page);
1113 			}
1114 			/* Get the $MFT page. */
1115 			mft_page = ntfs_map_page(vol->mft_ino->i_mapping,
1116 					index);
1117 			if (IS_ERR(mft_page)) {
1118 				ntfs_error(sb, "Failed to read $MFT.");
1119 				return false;
1120 			}
1121 			kmft = page_address(mft_page);
1122 			/* Get the $MFTMirr page. */
1123 			mirr_page = ntfs_map_page(vol->mftmirr_ino->i_mapping,
1124 					index);
1125 			if (IS_ERR(mirr_page)) {
1126 				ntfs_error(sb, "Failed to read $MFTMirr.");
1127 				goto mft_unmap_out;
1128 			}
1129 			kmirr = page_address(mirr_page);
1130 			++index;
1131 		}
1132 		/* Do not check the record if it is not in use. */
1133 		if (((MFT_RECORD*)kmft)->flags & MFT_RECORD_IN_USE) {
1134 			/* Make sure the record is ok. */
1135 			if (ntfs_is_baad_recordp((le32*)kmft)) {
1136 				ntfs_error(sb, "Incomplete multi sector "
1137 						"transfer detected in mft "
1138 						"record %i.", i);
1139 mm_unmap_out:
1140 				ntfs_unmap_page(mirr_page);
1141 mft_unmap_out:
1142 				ntfs_unmap_page(mft_page);
1143 				return false;
1144 			}
1145 		}
1146 		/* Do not check the mirror record if it is not in use. */
1147 		if (((MFT_RECORD*)kmirr)->flags & MFT_RECORD_IN_USE) {
1148 			if (ntfs_is_baad_recordp((le32*)kmirr)) {
1149 				ntfs_error(sb, "Incomplete multi sector "
1150 						"transfer detected in mft "
1151 						"mirror record %i.", i);
1152 				goto mm_unmap_out;
1153 			}
1154 		}
1155 		/* Get the amount of data in the current record. */
1156 		bytes = le32_to_cpu(((MFT_RECORD*)kmft)->bytes_in_use);
1157 		if (bytes < sizeof(MFT_RECORD_OLD) ||
1158 				bytes > vol->mft_record_size ||
1159 				ntfs_is_baad_recordp((le32*)kmft)) {
1160 			bytes = le32_to_cpu(((MFT_RECORD*)kmirr)->bytes_in_use);
1161 			if (bytes < sizeof(MFT_RECORD_OLD) ||
1162 					bytes > vol->mft_record_size ||
1163 					ntfs_is_baad_recordp((le32*)kmirr))
1164 				bytes = vol->mft_record_size;
1165 		}
1166 		/* Compare the two records. */
1167 		if (memcmp(kmft, kmirr, bytes)) {
1168 			ntfs_error(sb, "$MFT and $MFTMirr (record %i) do not "
1169 					"match.  Run ntfsfix or chkdsk.", i);
1170 			goto mm_unmap_out;
1171 		}
1172 		kmft += vol->mft_record_size;
1173 		kmirr += vol->mft_record_size;
1174 	} while (++i < vol->mftmirr_size);
1175 	/* Release the last pages. */
1176 	ntfs_unmap_page(mft_page);
1177 	ntfs_unmap_page(mirr_page);
1178 
1179 	/* Construct the mft mirror runlist by hand. */
1180 	rl2[0].vcn = 0;
1181 	rl2[0].lcn = vol->mftmirr_lcn;
1182 	rl2[0].length = (vol->mftmirr_size * vol->mft_record_size +
1183 			vol->cluster_size - 1) / vol->cluster_size;
1184 	rl2[1].vcn = rl2[0].length;
1185 	rl2[1].lcn = LCN_ENOENT;
1186 	rl2[1].length = 0;
1187 	/*
1188 	 * Because we have just read all of the mft mirror, we know we have
1189 	 * mapped the full runlist for it.
1190 	 */
1191 	mirr_ni = NTFS_I(vol->mftmirr_ino);
1192 	down_read(&mirr_ni->runlist.lock);
1193 	rl = mirr_ni->runlist.rl;
1194 	/* Compare the two runlists.  They must be identical. */
1195 	i = 0;
1196 	do {
1197 		if (rl2[i].vcn != rl[i].vcn || rl2[i].lcn != rl[i].lcn ||
1198 				rl2[i].length != rl[i].length) {
1199 			ntfs_error(sb, "$MFTMirr location mismatch.  "
1200 					"Run chkdsk.");
1201 			up_read(&mirr_ni->runlist.lock);
1202 			return false;
1203 		}
1204 	} while (rl2[i++].length);
1205 	up_read(&mirr_ni->runlist.lock);
1206 	ntfs_debug("Done.");
1207 	return true;
1208 }
1209 
1210 /**
1211  * load_and_check_logfile - load and check the logfile inode for a volume
1212  * @vol:	ntfs super block describing device whose logfile to load
1213  *
1214  * Return 'true' on success or 'false' on error.
1215  */
1216 static bool load_and_check_logfile(ntfs_volume *vol,
1217 		RESTART_PAGE_HEADER **rp)
1218 {
1219 	struct inode *tmp_ino;
1220 
1221 	ntfs_debug("Entering.");
1222 	tmp_ino = ntfs_iget(vol->sb, FILE_LogFile);
1223 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1224 		if (!IS_ERR(tmp_ino))
1225 			iput(tmp_ino);
1226 		/* Caller will display error message. */
1227 		return false;
1228 	}
1229 	if (!ntfs_check_logfile(tmp_ino, rp)) {
1230 		iput(tmp_ino);
1231 		/* ntfs_check_logfile() will have displayed error output. */
1232 		return false;
1233 	}
1234 	NInoSetSparseDisabled(NTFS_I(tmp_ino));
1235 	vol->logfile_ino = tmp_ino;
1236 	ntfs_debug("Done.");
1237 	return true;
1238 }
1239 
1240 #define NTFS_HIBERFIL_HEADER_SIZE	4096
1241 
1242 /**
1243  * check_windows_hibernation_status - check if Windows is suspended on a volume
1244  * @vol:	ntfs super block of device to check
1245  *
1246  * Check if Windows is hibernated on the ntfs volume @vol.  This is done by
1247  * looking for the file hiberfil.sys in the root directory of the volume.  If
1248  * the file is not present Windows is definitely not suspended.
1249  *
1250  * If hiberfil.sys exists and is less than 4kiB in size it means Windows is
1251  * definitely suspended (this volume is not the system volume).  Caveat:  on a
1252  * system with many volumes it is possible that the < 4kiB check is bogus but
1253  * for now this should do fine.
1254  *
1255  * If hiberfil.sys exists and is larger than 4kiB in size, we need to read the
1256  * hiberfil header (which is the first 4kiB).  If this begins with "hibr",
1257  * Windows is definitely suspended.  If it is completely full of zeroes,
1258  * Windows is definitely not hibernated.  Any other case is treated as if
1259  * Windows is suspended.  This caters for the above mentioned caveat of a
1260  * system with many volumes where no "hibr" magic would be present and there is
1261  * no zero header.
1262  *
1263  * Return 0 if Windows is not hibernated on the volume, >0 if Windows is
1264  * hibernated on the volume, and -errno on error.
1265  */
1266 static int check_windows_hibernation_status(ntfs_volume *vol)
1267 {
1268 	MFT_REF mref;
1269 	struct inode *vi;
1270 	struct page *page;
1271 	u32 *kaddr, *kend;
1272 	ntfs_name *name = NULL;
1273 	int ret = 1;
1274 	static const ntfschar hiberfil[13] = { cpu_to_le16('h'),
1275 			cpu_to_le16('i'), cpu_to_le16('b'),
1276 			cpu_to_le16('e'), cpu_to_le16('r'),
1277 			cpu_to_le16('f'), cpu_to_le16('i'),
1278 			cpu_to_le16('l'), cpu_to_le16('.'),
1279 			cpu_to_le16('s'), cpu_to_le16('y'),
1280 			cpu_to_le16('s'), 0 };
1281 
1282 	ntfs_debug("Entering.");
1283 	/*
1284 	 * Find the inode number for the hibernation file by looking up the
1285 	 * filename hiberfil.sys in the root directory.
1286 	 */
1287 	mutex_lock(&vol->root_ino->i_mutex);
1288 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->root_ino), hiberfil, 12,
1289 			&name);
1290 	mutex_unlock(&vol->root_ino->i_mutex);
1291 	if (IS_ERR_MREF(mref)) {
1292 		ret = MREF_ERR(mref);
1293 		/* If the file does not exist, Windows is not hibernated. */
1294 		if (ret == -ENOENT) {
1295 			ntfs_debug("hiberfil.sys not present.  Windows is not "
1296 					"hibernated on the volume.");
1297 			return 0;
1298 		}
1299 		/* A real error occurred. */
1300 		ntfs_error(vol->sb, "Failed to find inode number for "
1301 				"hiberfil.sys.");
1302 		return ret;
1303 	}
1304 	/* We do not care for the type of match that was found. */
1305 	kfree(name);
1306 	/* Get the inode. */
1307 	vi = ntfs_iget(vol->sb, MREF(mref));
1308 	if (IS_ERR(vi) || is_bad_inode(vi)) {
1309 		if (!IS_ERR(vi))
1310 			iput(vi);
1311 		ntfs_error(vol->sb, "Failed to load hiberfil.sys.");
1312 		return IS_ERR(vi) ? PTR_ERR(vi) : -EIO;
1313 	}
1314 	if (unlikely(i_size_read(vi) < NTFS_HIBERFIL_HEADER_SIZE)) {
1315 		ntfs_debug("hiberfil.sys is smaller than 4kiB (0x%llx).  "
1316 				"Windows is hibernated on the volume.  This "
1317 				"is not the system volume.", i_size_read(vi));
1318 		goto iput_out;
1319 	}
1320 	page = ntfs_map_page(vi->i_mapping, 0);
1321 	if (IS_ERR(page)) {
1322 		ntfs_error(vol->sb, "Failed to read from hiberfil.sys.");
1323 		ret = PTR_ERR(page);
1324 		goto iput_out;
1325 	}
1326 	kaddr = (u32*)page_address(page);
1327 	if (*(le32*)kaddr == cpu_to_le32(0x72626968)/*'hibr'*/) {
1328 		ntfs_debug("Magic \"hibr\" found in hiberfil.sys.  Windows is "
1329 				"hibernated on the volume.  This is the "
1330 				"system volume.");
1331 		goto unm_iput_out;
1332 	}
1333 	kend = kaddr + NTFS_HIBERFIL_HEADER_SIZE/sizeof(*kaddr);
1334 	do {
1335 		if (unlikely(*kaddr)) {
1336 			ntfs_debug("hiberfil.sys is larger than 4kiB "
1337 					"(0x%llx), does not contain the "
1338 					"\"hibr\" magic, and does not have a "
1339 					"zero header.  Windows is hibernated "
1340 					"on the volume.  This is not the "
1341 					"system volume.", i_size_read(vi));
1342 			goto unm_iput_out;
1343 		}
1344 	} while (++kaddr < kend);
1345 	ntfs_debug("hiberfil.sys contains a zero header.  Windows is not "
1346 			"hibernated on the volume.  This is the system "
1347 			"volume.");
1348 	ret = 0;
1349 unm_iput_out:
1350 	ntfs_unmap_page(page);
1351 iput_out:
1352 	iput(vi);
1353 	return ret;
1354 }
1355 
1356 /**
1357  * load_and_init_quota - load and setup the quota file for a volume if present
1358  * @vol:	ntfs super block describing device whose quota file to load
1359  *
1360  * Return 'true' on success or 'false' on error.  If $Quota is not present, we
1361  * leave vol->quota_ino as NULL and return success.
1362  */
1363 static bool load_and_init_quota(ntfs_volume *vol)
1364 {
1365 	MFT_REF mref;
1366 	struct inode *tmp_ino;
1367 	ntfs_name *name = NULL;
1368 	static const ntfschar Quota[7] = { cpu_to_le16('$'),
1369 			cpu_to_le16('Q'), cpu_to_le16('u'),
1370 			cpu_to_le16('o'), cpu_to_le16('t'),
1371 			cpu_to_le16('a'), 0 };
1372 	static ntfschar Q[3] = { cpu_to_le16('$'),
1373 			cpu_to_le16('Q'), 0 };
1374 
1375 	ntfs_debug("Entering.");
1376 	/*
1377 	 * Find the inode number for the quota file by looking up the filename
1378 	 * $Quota in the extended system files directory $Extend.
1379 	 */
1380 	mutex_lock(&vol->extend_ino->i_mutex);
1381 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), Quota, 6,
1382 			&name);
1383 	mutex_unlock(&vol->extend_ino->i_mutex);
1384 	if (IS_ERR_MREF(mref)) {
1385 		/*
1386 		 * If the file does not exist, quotas are disabled and have
1387 		 * never been enabled on this volume, just return success.
1388 		 */
1389 		if (MREF_ERR(mref) == -ENOENT) {
1390 			ntfs_debug("$Quota not present.  Volume does not have "
1391 					"quotas enabled.");
1392 			/*
1393 			 * No need to try to set quotas out of date if they are
1394 			 * not enabled.
1395 			 */
1396 			NVolSetQuotaOutOfDate(vol);
1397 			return true;
1398 		}
1399 		/* A real error occurred. */
1400 		ntfs_error(vol->sb, "Failed to find inode number for $Quota.");
1401 		return false;
1402 	}
1403 	/* We do not care for the type of match that was found. */
1404 	kfree(name);
1405 	/* Get the inode. */
1406 	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1407 	if (IS_ERR(tmp_ino) || is_bad_inode(tmp_ino)) {
1408 		if (!IS_ERR(tmp_ino))
1409 			iput(tmp_ino);
1410 		ntfs_error(vol->sb, "Failed to load $Quota.");
1411 		return false;
1412 	}
1413 	vol->quota_ino = tmp_ino;
1414 	/* Get the $Q index allocation attribute. */
1415 	tmp_ino = ntfs_index_iget(vol->quota_ino, Q, 2);
1416 	if (IS_ERR(tmp_ino)) {
1417 		ntfs_error(vol->sb, "Failed to load $Quota/$Q index.");
1418 		return false;
1419 	}
1420 	vol->quota_q_ino = tmp_ino;
1421 	ntfs_debug("Done.");
1422 	return true;
1423 }
1424 
1425 /**
1426  * load_and_init_usnjrnl - load and setup the transaction log if present
1427  * @vol:	ntfs super block describing device whose usnjrnl file to load
1428  *
1429  * Return 'true' on success or 'false' on error.
1430  *
1431  * If $UsnJrnl is not present or in the process of being disabled, we set
1432  * NVolUsnJrnlStamped() and return success.
1433  *
1434  * If the $UsnJrnl $DATA/$J attribute has a size equal to the lowest valid usn,
1435  * i.e. transaction logging has only just been enabled or the journal has been
1436  * stamped and nothing has been logged since, we also set NVolUsnJrnlStamped()
1437  * and return success.
1438  */
1439 static bool load_and_init_usnjrnl(ntfs_volume *vol)
1440 {
1441 	MFT_REF mref;
1442 	struct inode *tmp_ino;
1443 	ntfs_inode *tmp_ni;
1444 	struct page *page;
1445 	ntfs_name *name = NULL;
1446 	USN_HEADER *uh;
1447 	static const ntfschar UsnJrnl[9] = { cpu_to_le16('$'),
1448 			cpu_to_le16('U'), cpu_to_le16('s'),
1449 			cpu_to_le16('n'), cpu_to_le16('J'),
1450 			cpu_to_le16('r'), cpu_to_le16('n'),
1451 			cpu_to_le16('l'), 0 };
1452 	static ntfschar Max[5] = { cpu_to_le16('$'),
1453 			cpu_to_le16('M'), cpu_to_le16('a'),
1454 			cpu_to_le16('x'), 0 };
1455 	static ntfschar J[3] = { cpu_to_le16('$'),
1456 			cpu_to_le16('J'), 0 };
1457 
1458 	ntfs_debug("Entering.");
1459 	/*
1460 	 * Find the inode number for the transaction log file by looking up the
1461 	 * filename $UsnJrnl in the extended system files directory $Extend.
1462 	 */
1463 	mutex_lock(&vol->extend_ino->i_mutex);
1464 	mref = ntfs_lookup_inode_by_name(NTFS_I(vol->extend_ino), UsnJrnl, 8,
1465 			&name);
1466 	mutex_unlock(&vol->extend_ino->i_mutex);
1467 	if (IS_ERR_MREF(mref)) {
1468 		/*
1469 		 * If the file does not exist, transaction logging is disabled,
1470 		 * just return success.
1471 		 */
1472 		if (MREF_ERR(mref) == -ENOENT) {
1473 			ntfs_debug("$UsnJrnl not present.  Volume does not "
1474 					"have transaction logging enabled.");
1475 not_enabled:
1476 			/*
1477 			 * No need to try to stamp the transaction log if
1478 			 * transaction logging is not enabled.
1479 			 */
1480 			NVolSetUsnJrnlStamped(vol);
1481 			return true;
1482 		}
1483 		/* A real error occurred. */
1484 		ntfs_error(vol->sb, "Failed to find inode number for "
1485 				"$UsnJrnl.");
1486 		return false;
1487 	}
1488 	/* We do not care for the type of match that was found. */
1489 	kfree(name);
1490 	/* Get the inode. */
1491 	tmp_ino = ntfs_iget(vol->sb, MREF(mref));
1492 	if (unlikely(IS_ERR(tmp_ino) || is_bad_inode(tmp_ino))) {
1493 		if (!IS_ERR(tmp_ino))
1494 			iput(tmp_ino);
1495 		ntfs_error(vol->sb, "Failed to load $UsnJrnl.");
1496 		return false;
1497 	}
1498 	vol->usnjrnl_ino = tmp_ino;
1499 	/*
1500 	 * If the transaction log is in the process of being deleted, we can
1501 	 * ignore it.
1502 	 */
1503 	if (unlikely(vol->vol_flags & VOLUME_DELETE_USN_UNDERWAY)) {
1504 		ntfs_debug("$UsnJrnl in the process of being disabled.  "
1505 				"Volume does not have transaction logging "
1506 				"enabled.");
1507 		goto not_enabled;
1508 	}
1509 	/* Get the $DATA/$Max attribute. */
1510 	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, Max, 4);
1511 	if (IS_ERR(tmp_ino)) {
1512 		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$Max "
1513 				"attribute.");
1514 		return false;
1515 	}
1516 	vol->usnjrnl_max_ino = tmp_ino;
1517 	if (unlikely(i_size_read(tmp_ino) < sizeof(USN_HEADER))) {
1518 		ntfs_error(vol->sb, "Found corrupt $UsnJrnl/$DATA/$Max "
1519 				"attribute (size is 0x%llx but should be at "
1520 				"least 0x%zx bytes).", i_size_read(tmp_ino),
1521 				sizeof(USN_HEADER));
1522 		return false;
1523 	}
1524 	/* Get the $DATA/$J attribute. */
1525 	tmp_ino = ntfs_attr_iget(vol->usnjrnl_ino, AT_DATA, J, 2);
1526 	if (IS_ERR(tmp_ino)) {
1527 		ntfs_error(vol->sb, "Failed to load $UsnJrnl/$DATA/$J "
1528 				"attribute.");
1529 		return false;
1530 	}
1531 	vol->usnjrnl_j_ino = tmp_ino;
1532 	/* Verify $J is non-resident and sparse. */
1533 	tmp_ni = NTFS_I(vol->usnjrnl_j_ino);
1534 	if (unlikely(!NInoNonResident(tmp_ni) || !NInoSparse(tmp_ni))) {
1535 		ntfs_error(vol->sb, "$UsnJrnl/$DATA/$J attribute is resident "
1536 				"and/or not sparse.");
1537 		return false;
1538 	}
1539 	/* Read the USN_HEADER from $DATA/$Max. */
1540 	page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
1541 	if (IS_ERR(page)) {
1542 		ntfs_error(vol->sb, "Failed to read from $UsnJrnl/$DATA/$Max "
1543 				"attribute.");
1544 		return false;
1545 	}
1546 	uh = (USN_HEADER*)page_address(page);
1547 	/* Sanity check the $Max. */
1548 	if (unlikely(sle64_to_cpu(uh->allocation_delta) >
1549 			sle64_to_cpu(uh->maximum_size))) {
1550 		ntfs_error(vol->sb, "Allocation delta (0x%llx) exceeds "
1551 				"maximum size (0x%llx).  $UsnJrnl is corrupt.",
1552 				(long long)sle64_to_cpu(uh->allocation_delta),
1553 				(long long)sle64_to_cpu(uh->maximum_size));
1554 		ntfs_unmap_page(page);
1555 		return false;
1556 	}
1557 	/*
1558 	 * If the transaction log has been stamped and nothing has been written
1559 	 * to it since, we do not need to stamp it.
1560 	 */
1561 	if (unlikely(sle64_to_cpu(uh->lowest_valid_usn) >=
1562 			i_size_read(vol->usnjrnl_j_ino))) {
1563 		if (likely(sle64_to_cpu(uh->lowest_valid_usn) ==
1564 				i_size_read(vol->usnjrnl_j_ino))) {
1565 			ntfs_unmap_page(page);
1566 			ntfs_debug("$UsnJrnl is enabled but nothing has been "
1567 					"logged since it was last stamped.  "
1568 					"Treating this as if the volume does "
1569 					"not have transaction logging "
1570 					"enabled.");
1571 			goto not_enabled;
1572 		}
1573 		ntfs_error(vol->sb, "$UsnJrnl has lowest valid usn (0x%llx) "
1574 				"which is out of bounds (0x%llx).  $UsnJrnl "
1575 				"is corrupt.",
1576 				(long long)sle64_to_cpu(uh->lowest_valid_usn),
1577 				i_size_read(vol->usnjrnl_j_ino));
1578 		ntfs_unmap_page(page);
1579 		return false;
1580 	}
1581 	ntfs_unmap_page(page);
1582 	ntfs_debug("Done.");
1583 	return true;
1584 }
1585 
1586 /**
1587  * load_and_init_attrdef - load the attribute definitions table for a volume
1588  * @vol:	ntfs super block describing device whose attrdef to load
1589  *
1590  * Return 'true' on success or 'false' on error.
1591  */
1592 static bool load_and_init_attrdef(ntfs_volume *vol)
1593 {
1594 	loff_t i_size;
1595 	struct super_block *sb = vol->sb;
1596 	struct inode *ino;
1597 	struct page *page;
1598 	pgoff_t index, max_index;
1599 	unsigned int size;
1600 
1601 	ntfs_debug("Entering.");
1602 	/* Read attrdef table and setup vol->attrdef and vol->attrdef_size. */
1603 	ino = ntfs_iget(sb, FILE_AttrDef);
1604 	if (IS_ERR(ino) || is_bad_inode(ino)) {
1605 		if (!IS_ERR(ino))
1606 			iput(ino);
1607 		goto failed;
1608 	}
1609 	NInoSetSparseDisabled(NTFS_I(ino));
1610 	/* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
1611 	i_size = i_size_read(ino);
1612 	if (i_size <= 0 || i_size > 0x7fffffff)
1613 		goto iput_failed;
1614 	vol->attrdef = (ATTR_DEF*)ntfs_malloc_nofs(i_size);
1615 	if (!vol->attrdef)
1616 		goto iput_failed;
1617 	index = 0;
1618 	max_index = i_size >> PAGE_CACHE_SHIFT;
1619 	size = PAGE_CACHE_SIZE;
1620 	while (index < max_index) {
1621 		/* Read the attrdef table and copy it into the linear buffer. */
1622 read_partial_attrdef_page:
1623 		page = ntfs_map_page(ino->i_mapping, index);
1624 		if (IS_ERR(page))
1625 			goto free_iput_failed;
1626 		memcpy((u8*)vol->attrdef + (index++ << PAGE_CACHE_SHIFT),
1627 				page_address(page), size);
1628 		ntfs_unmap_page(page);
1629 	};
1630 	if (size == PAGE_CACHE_SIZE) {
1631 		size = i_size & ~PAGE_CACHE_MASK;
1632 		if (size)
1633 			goto read_partial_attrdef_page;
1634 	}
1635 	vol->attrdef_size = i_size;
1636 	ntfs_debug("Read %llu bytes from $AttrDef.", i_size);
1637 	iput(ino);
1638 	return true;
1639 free_iput_failed:
1640 	ntfs_free(vol->attrdef);
1641 	vol->attrdef = NULL;
1642 iput_failed:
1643 	iput(ino);
1644 failed:
1645 	ntfs_error(sb, "Failed to initialize attribute definition table.");
1646 	return false;
1647 }
1648 
1649 #endif /* NTFS_RW */
1650 
1651 /**
1652  * load_and_init_upcase - load the upcase table for an ntfs volume
1653  * @vol:	ntfs super block describing device whose upcase to load
1654  *
1655  * Return 'true' on success or 'false' on error.
1656  */
1657 static bool load_and_init_upcase(ntfs_volume *vol)
1658 {
1659 	loff_t i_size;
1660 	struct super_block *sb = vol->sb;
1661 	struct inode *ino;
1662 	struct page *page;
1663 	pgoff_t index, max_index;
1664 	unsigned int size;
1665 	int i, max;
1666 
1667 	ntfs_debug("Entering.");
1668 	/* Read upcase table and setup vol->upcase and vol->upcase_len. */
1669 	ino = ntfs_iget(sb, FILE_UpCase);
1670 	if (IS_ERR(ino) || is_bad_inode(ino)) {
1671 		if (!IS_ERR(ino))
1672 			iput(ino);
1673 		goto upcase_failed;
1674 	}
1675 	/*
1676 	 * The upcase size must not be above 64k Unicode characters, must not
1677 	 * be zero and must be a multiple of sizeof(ntfschar).
1678 	 */
1679 	i_size = i_size_read(ino);
1680 	if (!i_size || i_size & (sizeof(ntfschar) - 1) ||
1681 			i_size > 64ULL * 1024 * sizeof(ntfschar))
1682 		goto iput_upcase_failed;
1683 	vol->upcase = (ntfschar*)ntfs_malloc_nofs(i_size);
1684 	if (!vol->upcase)
1685 		goto iput_upcase_failed;
1686 	index = 0;
1687 	max_index = i_size >> PAGE_CACHE_SHIFT;
1688 	size = PAGE_CACHE_SIZE;
1689 	while (index < max_index) {
1690 		/* Read the upcase table and copy it into the linear buffer. */
1691 read_partial_upcase_page:
1692 		page = ntfs_map_page(ino->i_mapping, index);
1693 		if (IS_ERR(page))
1694 			goto iput_upcase_failed;
1695 		memcpy((char*)vol->upcase + (index++ << PAGE_CACHE_SHIFT),
1696 				page_address(page), size);
1697 		ntfs_unmap_page(page);
1698 	};
1699 	if (size == PAGE_CACHE_SIZE) {
1700 		size = i_size & ~PAGE_CACHE_MASK;
1701 		if (size)
1702 			goto read_partial_upcase_page;
1703 	}
1704 	vol->upcase_len = i_size >> UCHAR_T_SIZE_BITS;
1705 	ntfs_debug("Read %llu bytes from $UpCase (expected %zu bytes).",
1706 			i_size, 64 * 1024 * sizeof(ntfschar));
1707 	iput(ino);
1708 	mutex_lock(&ntfs_lock);
1709 	if (!default_upcase) {
1710 		ntfs_debug("Using volume specified $UpCase since default is "
1711 				"not present.");
1712 		mutex_unlock(&ntfs_lock);
1713 		return true;
1714 	}
1715 	max = default_upcase_len;
1716 	if (max > vol->upcase_len)
1717 		max = vol->upcase_len;
1718 	for (i = 0; i < max; i++)
1719 		if (vol->upcase[i] != default_upcase[i])
1720 			break;
1721 	if (i == max) {
1722 		ntfs_free(vol->upcase);
1723 		vol->upcase = default_upcase;
1724 		vol->upcase_len = max;
1725 		ntfs_nr_upcase_users++;
1726 		mutex_unlock(&ntfs_lock);
1727 		ntfs_debug("Volume specified $UpCase matches default. Using "
1728 				"default.");
1729 		return true;
1730 	}
1731 	mutex_unlock(&ntfs_lock);
1732 	ntfs_debug("Using volume specified $UpCase since it does not match "
1733 			"the default.");
1734 	return true;
1735 iput_upcase_failed:
1736 	iput(ino);
1737 	ntfs_free(vol->upcase);
1738 	vol->upcase = NULL;
1739 upcase_failed:
1740 	mutex_lock(&ntfs_lock);
1741 	if (default_upcase) {
1742 		vol->upcase = default_upcase;
1743 		vol->upcase_len = default_upcase_len;
1744 		ntfs_nr_upcase_users++;
1745 		mutex_unlock(&ntfs_lock);
1746 		ntfs_error(sb, "Failed to load $UpCase from the volume. Using "
1747 				"default.");
1748 		return true;
1749 	}
1750 	mutex_unlock(&ntfs_lock);
1751 	ntfs_error(sb, "Failed to initialize upcase table.");
1752 	return false;
1753 }
1754 
1755 /*
1756  * The lcn and mft bitmap inodes are NTFS-internal inodes with
1757  * their own special locking rules:
1758  */
1759 static struct lock_class_key
1760 	lcnbmp_runlist_lock_key, lcnbmp_mrec_lock_key,
1761 	mftbmp_runlist_lock_key, mftbmp_mrec_lock_key;
1762 
1763 /**
1764  * load_system_files - open the system files using normal functions
1765  * @vol:	ntfs super block describing device whose system files to load
1766  *
1767  * Open the system files with normal access functions and complete setting up
1768  * the ntfs super block @vol.
1769  *
1770  * Return 'true' on success or 'false' on error.
1771  */
1772 static bool load_system_files(ntfs_volume *vol)
1773 {
1774 	struct super_block *sb = vol->sb;
1775 	MFT_RECORD *m;
1776 	VOLUME_INFORMATION *vi;
1777 	ntfs_attr_search_ctx *ctx;
1778 #ifdef NTFS_RW
1779 	RESTART_PAGE_HEADER *rp;
1780 	int err;
1781 #endif /* NTFS_RW */
1782 
1783 	ntfs_debug("Entering.");
1784 #ifdef NTFS_RW
1785 	/* Get mft mirror inode compare the contents of $MFT and $MFTMirr. */
1786 	if (!load_and_init_mft_mirror(vol) || !check_mft_mirror(vol)) {
1787 		static const char *es1 = "Failed to load $MFTMirr";
1788 		static const char *es2 = "$MFTMirr does not match $MFT";
1789 		static const char *es3 = ".  Run ntfsfix and/or chkdsk.";
1790 
1791 		/* If a read-write mount, convert it to a read-only mount. */
1792 		if (!(sb->s_flags & MS_RDONLY)) {
1793 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1794 					ON_ERRORS_CONTINUE))) {
1795 				ntfs_error(sb, "%s and neither on_errors="
1796 						"continue nor on_errors="
1797 						"remount-ro was specified%s",
1798 						!vol->mftmirr_ino ? es1 : es2,
1799 						es3);
1800 				goto iput_mirr_err_out;
1801 			}
1802 			sb->s_flags |= MS_RDONLY;
1803 			ntfs_error(sb, "%s.  Mounting read-only%s",
1804 					!vol->mftmirr_ino ? es1 : es2, es3);
1805 		} else
1806 			ntfs_warning(sb, "%s.  Will not be able to remount "
1807 					"read-write%s",
1808 					!vol->mftmirr_ino ? es1 : es2, es3);
1809 		/* This will prevent a read-write remount. */
1810 		NVolSetErrors(vol);
1811 	}
1812 #endif /* NTFS_RW */
1813 	/* Get mft bitmap attribute inode. */
1814 	vol->mftbmp_ino = ntfs_attr_iget(vol->mft_ino, AT_BITMAP, NULL, 0);
1815 	if (IS_ERR(vol->mftbmp_ino)) {
1816 		ntfs_error(sb, "Failed to load $MFT/$BITMAP attribute.");
1817 		goto iput_mirr_err_out;
1818 	}
1819 	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->runlist.lock,
1820 			   &mftbmp_runlist_lock_key);
1821 	lockdep_set_class(&NTFS_I(vol->mftbmp_ino)->mrec_lock,
1822 			   &mftbmp_mrec_lock_key);
1823 	/* Read upcase table and setup @vol->upcase and @vol->upcase_len. */
1824 	if (!load_and_init_upcase(vol))
1825 		goto iput_mftbmp_err_out;
1826 #ifdef NTFS_RW
1827 	/*
1828 	 * Read attribute definitions table and setup @vol->attrdef and
1829 	 * @vol->attrdef_size.
1830 	 */
1831 	if (!load_and_init_attrdef(vol))
1832 		goto iput_upcase_err_out;
1833 #endif /* NTFS_RW */
1834 	/*
1835 	 * Get the cluster allocation bitmap inode and verify the size, no
1836 	 * need for any locking at this stage as we are already running
1837 	 * exclusively as we are mount in progress task.
1838 	 */
1839 	vol->lcnbmp_ino = ntfs_iget(sb, FILE_Bitmap);
1840 	if (IS_ERR(vol->lcnbmp_ino) || is_bad_inode(vol->lcnbmp_ino)) {
1841 		if (!IS_ERR(vol->lcnbmp_ino))
1842 			iput(vol->lcnbmp_ino);
1843 		goto bitmap_failed;
1844 	}
1845 	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->runlist.lock,
1846 			   &lcnbmp_runlist_lock_key);
1847 	lockdep_set_class(&NTFS_I(vol->lcnbmp_ino)->mrec_lock,
1848 			   &lcnbmp_mrec_lock_key);
1849 
1850 	NInoSetSparseDisabled(NTFS_I(vol->lcnbmp_ino));
1851 	if ((vol->nr_clusters + 7) >> 3 > i_size_read(vol->lcnbmp_ino)) {
1852 		iput(vol->lcnbmp_ino);
1853 bitmap_failed:
1854 		ntfs_error(sb, "Failed to load $Bitmap.");
1855 		goto iput_attrdef_err_out;
1856 	}
1857 	/*
1858 	 * Get the volume inode and setup our cache of the volume flags and
1859 	 * version.
1860 	 */
1861 	vol->vol_ino = ntfs_iget(sb, FILE_Volume);
1862 	if (IS_ERR(vol->vol_ino) || is_bad_inode(vol->vol_ino)) {
1863 		if (!IS_ERR(vol->vol_ino))
1864 			iput(vol->vol_ino);
1865 volume_failed:
1866 		ntfs_error(sb, "Failed to load $Volume.");
1867 		goto iput_lcnbmp_err_out;
1868 	}
1869 	m = map_mft_record(NTFS_I(vol->vol_ino));
1870 	if (IS_ERR(m)) {
1871 iput_volume_failed:
1872 		iput(vol->vol_ino);
1873 		goto volume_failed;
1874 	}
1875 	if (!(ctx = ntfs_attr_get_search_ctx(NTFS_I(vol->vol_ino), m))) {
1876 		ntfs_error(sb, "Failed to get attribute search context.");
1877 		goto get_ctx_vol_failed;
1878 	}
1879 	if (ntfs_attr_lookup(AT_VOLUME_INFORMATION, NULL, 0, 0, 0, NULL, 0,
1880 			ctx) || ctx->attr->non_resident || ctx->attr->flags) {
1881 err_put_vol:
1882 		ntfs_attr_put_search_ctx(ctx);
1883 get_ctx_vol_failed:
1884 		unmap_mft_record(NTFS_I(vol->vol_ino));
1885 		goto iput_volume_failed;
1886 	}
1887 	vi = (VOLUME_INFORMATION*)((char*)ctx->attr +
1888 			le16_to_cpu(ctx->attr->data.resident.value_offset));
1889 	/* Some bounds checks. */
1890 	if ((u8*)vi < (u8*)ctx->attr || (u8*)vi +
1891 			le32_to_cpu(ctx->attr->data.resident.value_length) >
1892 			(u8*)ctx->attr + le32_to_cpu(ctx->attr->length))
1893 		goto err_put_vol;
1894 	/* Copy the volume flags and version to the ntfs_volume structure. */
1895 	vol->vol_flags = vi->flags;
1896 	vol->major_ver = vi->major_ver;
1897 	vol->minor_ver = vi->minor_ver;
1898 	ntfs_attr_put_search_ctx(ctx);
1899 	unmap_mft_record(NTFS_I(vol->vol_ino));
1900 	pr_info("volume version %i.%i.\n", vol->major_ver,
1901 			vol->minor_ver);
1902 	if (vol->major_ver < 3 && NVolSparseEnabled(vol)) {
1903 		ntfs_warning(vol->sb, "Disabling sparse support due to NTFS "
1904 				"volume version %i.%i (need at least version "
1905 				"3.0).", vol->major_ver, vol->minor_ver);
1906 		NVolClearSparseEnabled(vol);
1907 	}
1908 #ifdef NTFS_RW
1909 	/* Make sure that no unsupported volume flags are set. */
1910 	if (vol->vol_flags & VOLUME_MUST_MOUNT_RO_MASK) {
1911 		static const char *es1a = "Volume is dirty";
1912 		static const char *es1b = "Volume has been modified by chkdsk";
1913 		static const char *es1c = "Volume has unsupported flags set";
1914 		static const char *es2a = ".  Run chkdsk and mount in Windows.";
1915 		static const char *es2b = ".  Mount in Windows.";
1916 		const char *es1, *es2;
1917 
1918 		es2 = es2a;
1919 		if (vol->vol_flags & VOLUME_IS_DIRTY)
1920 			es1 = es1a;
1921 		else if (vol->vol_flags & VOLUME_MODIFIED_BY_CHKDSK) {
1922 			es1 = es1b;
1923 			es2 = es2b;
1924 		} else {
1925 			es1 = es1c;
1926 			ntfs_warning(sb, "Unsupported volume flags 0x%x "
1927 					"encountered.",
1928 					(unsigned)le16_to_cpu(vol->vol_flags));
1929 		}
1930 		/* If a read-write mount, convert it to a read-only mount. */
1931 		if (!(sb->s_flags & MS_RDONLY)) {
1932 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1933 					ON_ERRORS_CONTINUE))) {
1934 				ntfs_error(sb, "%s and neither on_errors="
1935 						"continue nor on_errors="
1936 						"remount-ro was specified%s",
1937 						es1, es2);
1938 				goto iput_vol_err_out;
1939 			}
1940 			sb->s_flags |= MS_RDONLY;
1941 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1942 		} else
1943 			ntfs_warning(sb, "%s.  Will not be able to remount "
1944 					"read-write%s", es1, es2);
1945 		/*
1946 		 * Do not set NVolErrors() because ntfs_remount() re-checks the
1947 		 * flags which we need to do in case any flags have changed.
1948 		 */
1949 	}
1950 	/*
1951 	 * Get the inode for the logfile, check it and determine if the volume
1952 	 * was shutdown cleanly.
1953 	 */
1954 	rp = NULL;
1955 	if (!load_and_check_logfile(vol, &rp) ||
1956 			!ntfs_is_logfile_clean(vol->logfile_ino, rp)) {
1957 		static const char *es1a = "Failed to load $LogFile";
1958 		static const char *es1b = "$LogFile is not clean";
1959 		static const char *es2 = ".  Mount in Windows.";
1960 		const char *es1;
1961 
1962 		es1 = !vol->logfile_ino ? es1a : es1b;
1963 		/* If a read-write mount, convert it to a read-only mount. */
1964 		if (!(sb->s_flags & MS_RDONLY)) {
1965 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
1966 					ON_ERRORS_CONTINUE))) {
1967 				ntfs_error(sb, "%s and neither on_errors="
1968 						"continue nor on_errors="
1969 						"remount-ro was specified%s",
1970 						es1, es2);
1971 				if (vol->logfile_ino) {
1972 					BUG_ON(!rp);
1973 					ntfs_free(rp);
1974 				}
1975 				goto iput_logfile_err_out;
1976 			}
1977 			sb->s_flags |= MS_RDONLY;
1978 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
1979 		} else
1980 			ntfs_warning(sb, "%s.  Will not be able to remount "
1981 					"read-write%s", es1, es2);
1982 		/* This will prevent a read-write remount. */
1983 		NVolSetErrors(vol);
1984 	}
1985 	ntfs_free(rp);
1986 #endif /* NTFS_RW */
1987 	/* Get the root directory inode so we can do path lookups. */
1988 	vol->root_ino = ntfs_iget(sb, FILE_root);
1989 	if (IS_ERR(vol->root_ino) || is_bad_inode(vol->root_ino)) {
1990 		if (!IS_ERR(vol->root_ino))
1991 			iput(vol->root_ino);
1992 		ntfs_error(sb, "Failed to load root directory.");
1993 		goto iput_logfile_err_out;
1994 	}
1995 #ifdef NTFS_RW
1996 	/*
1997 	 * Check if Windows is suspended to disk on the target volume.  If it
1998 	 * is hibernated, we must not write *anything* to the disk so set
1999 	 * NVolErrors() without setting the dirty volume flag and mount
2000 	 * read-only.  This will prevent read-write remounting and it will also
2001 	 * prevent all writes.
2002 	 */
2003 	err = check_windows_hibernation_status(vol);
2004 	if (unlikely(err)) {
2005 		static const char *es1a = "Failed to determine if Windows is "
2006 				"hibernated";
2007 		static const char *es1b = "Windows is hibernated";
2008 		static const char *es2 = ".  Run chkdsk.";
2009 		const char *es1;
2010 
2011 		es1 = err < 0 ? es1a : es1b;
2012 		/* If a read-write mount, convert it to a read-only mount. */
2013 		if (!(sb->s_flags & MS_RDONLY)) {
2014 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2015 					ON_ERRORS_CONTINUE))) {
2016 				ntfs_error(sb, "%s and neither on_errors="
2017 						"continue nor on_errors="
2018 						"remount-ro was specified%s",
2019 						es1, es2);
2020 				goto iput_root_err_out;
2021 			}
2022 			sb->s_flags |= MS_RDONLY;
2023 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2024 		} else
2025 			ntfs_warning(sb, "%s.  Will not be able to remount "
2026 					"read-write%s", es1, es2);
2027 		/* This will prevent a read-write remount. */
2028 		NVolSetErrors(vol);
2029 	}
2030 	/* If (still) a read-write mount, mark the volume dirty. */
2031 	if (!(sb->s_flags & MS_RDONLY) &&
2032 			ntfs_set_volume_flags(vol, VOLUME_IS_DIRTY)) {
2033 		static const char *es1 = "Failed to set dirty bit in volume "
2034 				"information flags";
2035 		static const char *es2 = ".  Run chkdsk.";
2036 
2037 		/* Convert to a read-only mount. */
2038 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2039 				ON_ERRORS_CONTINUE))) {
2040 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2041 					"on_errors=remount-ro was specified%s",
2042 					es1, es2);
2043 			goto iput_root_err_out;
2044 		}
2045 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2046 		sb->s_flags |= MS_RDONLY;
2047 		/*
2048 		 * Do not set NVolErrors() because ntfs_remount() might manage
2049 		 * to set the dirty flag in which case all would be well.
2050 		 */
2051 	}
2052 #if 0
2053 	// TODO: Enable this code once we start modifying anything that is
2054 	//	 different between NTFS 1.2 and 3.x...
2055 	/*
2056 	 * If (still) a read-write mount, set the NT4 compatibility flag on
2057 	 * newer NTFS version volumes.
2058 	 */
2059 	if (!(sb->s_flags & MS_RDONLY) && (vol->major_ver > 1) &&
2060 			ntfs_set_volume_flags(vol, VOLUME_MOUNTED_ON_NT4)) {
2061 		static const char *es1 = "Failed to set NT4 compatibility flag";
2062 		static const char *es2 = ".  Run chkdsk.";
2063 
2064 		/* Convert to a read-only mount. */
2065 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2066 				ON_ERRORS_CONTINUE))) {
2067 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2068 					"on_errors=remount-ro was specified%s",
2069 					es1, es2);
2070 			goto iput_root_err_out;
2071 		}
2072 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2073 		sb->s_flags |= MS_RDONLY;
2074 		NVolSetErrors(vol);
2075 	}
2076 #endif
2077 	/* If (still) a read-write mount, empty the logfile. */
2078 	if (!(sb->s_flags & MS_RDONLY) &&
2079 			!ntfs_empty_logfile(vol->logfile_ino)) {
2080 		static const char *es1 = "Failed to empty $LogFile";
2081 		static const char *es2 = ".  Mount in Windows.";
2082 
2083 		/* Convert to a read-only mount. */
2084 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2085 				ON_ERRORS_CONTINUE))) {
2086 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2087 					"on_errors=remount-ro was specified%s",
2088 					es1, es2);
2089 			goto iput_root_err_out;
2090 		}
2091 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2092 		sb->s_flags |= MS_RDONLY;
2093 		NVolSetErrors(vol);
2094 	}
2095 #endif /* NTFS_RW */
2096 	/* If on NTFS versions before 3.0, we are done. */
2097 	if (unlikely(vol->major_ver < 3))
2098 		return true;
2099 	/* NTFS 3.0+ specific initialization. */
2100 	/* Get the security descriptors inode. */
2101 	vol->secure_ino = ntfs_iget(sb, FILE_Secure);
2102 	if (IS_ERR(vol->secure_ino) || is_bad_inode(vol->secure_ino)) {
2103 		if (!IS_ERR(vol->secure_ino))
2104 			iput(vol->secure_ino);
2105 		ntfs_error(sb, "Failed to load $Secure.");
2106 		goto iput_root_err_out;
2107 	}
2108 	// TODO: Initialize security.
2109 	/* Get the extended system files' directory inode. */
2110 	vol->extend_ino = ntfs_iget(sb, FILE_Extend);
2111 	if (IS_ERR(vol->extend_ino) || is_bad_inode(vol->extend_ino)) {
2112 		if (!IS_ERR(vol->extend_ino))
2113 			iput(vol->extend_ino);
2114 		ntfs_error(sb, "Failed to load $Extend.");
2115 		goto iput_sec_err_out;
2116 	}
2117 #ifdef NTFS_RW
2118 	/* Find the quota file, load it if present, and set it up. */
2119 	if (!load_and_init_quota(vol)) {
2120 		static const char *es1 = "Failed to load $Quota";
2121 		static const char *es2 = ".  Run chkdsk.";
2122 
2123 		/* If a read-write mount, convert it to a read-only mount. */
2124 		if (!(sb->s_flags & MS_RDONLY)) {
2125 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2126 					ON_ERRORS_CONTINUE))) {
2127 				ntfs_error(sb, "%s and neither on_errors="
2128 						"continue nor on_errors="
2129 						"remount-ro was specified%s",
2130 						es1, es2);
2131 				goto iput_quota_err_out;
2132 			}
2133 			sb->s_flags |= MS_RDONLY;
2134 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2135 		} else
2136 			ntfs_warning(sb, "%s.  Will not be able to remount "
2137 					"read-write%s", es1, es2);
2138 		/* This will prevent a read-write remount. */
2139 		NVolSetErrors(vol);
2140 	}
2141 	/* If (still) a read-write mount, mark the quotas out of date. */
2142 	if (!(sb->s_flags & MS_RDONLY) &&
2143 			!ntfs_mark_quotas_out_of_date(vol)) {
2144 		static const char *es1 = "Failed to mark quotas out of date";
2145 		static const char *es2 = ".  Run chkdsk.";
2146 
2147 		/* Convert to a read-only mount. */
2148 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2149 				ON_ERRORS_CONTINUE))) {
2150 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2151 					"on_errors=remount-ro was specified%s",
2152 					es1, es2);
2153 			goto iput_quota_err_out;
2154 		}
2155 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2156 		sb->s_flags |= MS_RDONLY;
2157 		NVolSetErrors(vol);
2158 	}
2159 	/*
2160 	 * Find the transaction log file ($UsnJrnl), load it if present, check
2161 	 * it, and set it up.
2162 	 */
2163 	if (!load_and_init_usnjrnl(vol)) {
2164 		static const char *es1 = "Failed to load $UsnJrnl";
2165 		static const char *es2 = ".  Run chkdsk.";
2166 
2167 		/* If a read-write mount, convert it to a read-only mount. */
2168 		if (!(sb->s_flags & MS_RDONLY)) {
2169 			if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2170 					ON_ERRORS_CONTINUE))) {
2171 				ntfs_error(sb, "%s and neither on_errors="
2172 						"continue nor on_errors="
2173 						"remount-ro was specified%s",
2174 						es1, es2);
2175 				goto iput_usnjrnl_err_out;
2176 			}
2177 			sb->s_flags |= MS_RDONLY;
2178 			ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2179 		} else
2180 			ntfs_warning(sb, "%s.  Will not be able to remount "
2181 					"read-write%s", es1, es2);
2182 		/* This will prevent a read-write remount. */
2183 		NVolSetErrors(vol);
2184 	}
2185 	/* If (still) a read-write mount, stamp the transaction log. */
2186 	if (!(sb->s_flags & MS_RDONLY) && !ntfs_stamp_usnjrnl(vol)) {
2187 		static const char *es1 = "Failed to stamp transaction log "
2188 				"($UsnJrnl)";
2189 		static const char *es2 = ".  Run chkdsk.";
2190 
2191 		/* Convert to a read-only mount. */
2192 		if (!(vol->on_errors & (ON_ERRORS_REMOUNT_RO |
2193 				ON_ERRORS_CONTINUE))) {
2194 			ntfs_error(sb, "%s and neither on_errors=continue nor "
2195 					"on_errors=remount-ro was specified%s",
2196 					es1, es2);
2197 			goto iput_usnjrnl_err_out;
2198 		}
2199 		ntfs_error(sb, "%s.  Mounting read-only%s", es1, es2);
2200 		sb->s_flags |= MS_RDONLY;
2201 		NVolSetErrors(vol);
2202 	}
2203 #endif /* NTFS_RW */
2204 	return true;
2205 #ifdef NTFS_RW
2206 iput_usnjrnl_err_out:
2207 	if (vol->usnjrnl_j_ino)
2208 		iput(vol->usnjrnl_j_ino);
2209 	if (vol->usnjrnl_max_ino)
2210 		iput(vol->usnjrnl_max_ino);
2211 	if (vol->usnjrnl_ino)
2212 		iput(vol->usnjrnl_ino);
2213 iput_quota_err_out:
2214 	if (vol->quota_q_ino)
2215 		iput(vol->quota_q_ino);
2216 	if (vol->quota_ino)
2217 		iput(vol->quota_ino);
2218 	iput(vol->extend_ino);
2219 #endif /* NTFS_RW */
2220 iput_sec_err_out:
2221 	iput(vol->secure_ino);
2222 iput_root_err_out:
2223 	iput(vol->root_ino);
2224 iput_logfile_err_out:
2225 #ifdef NTFS_RW
2226 	if (vol->logfile_ino)
2227 		iput(vol->logfile_ino);
2228 iput_vol_err_out:
2229 #endif /* NTFS_RW */
2230 	iput(vol->vol_ino);
2231 iput_lcnbmp_err_out:
2232 	iput(vol->lcnbmp_ino);
2233 iput_attrdef_err_out:
2234 	vol->attrdef_size = 0;
2235 	if (vol->attrdef) {
2236 		ntfs_free(vol->attrdef);
2237 		vol->attrdef = NULL;
2238 	}
2239 #ifdef NTFS_RW
2240 iput_upcase_err_out:
2241 #endif /* NTFS_RW */
2242 	vol->upcase_len = 0;
2243 	mutex_lock(&ntfs_lock);
2244 	if (vol->upcase == default_upcase) {
2245 		ntfs_nr_upcase_users--;
2246 		vol->upcase = NULL;
2247 	}
2248 	mutex_unlock(&ntfs_lock);
2249 	if (vol->upcase) {
2250 		ntfs_free(vol->upcase);
2251 		vol->upcase = NULL;
2252 	}
2253 iput_mftbmp_err_out:
2254 	iput(vol->mftbmp_ino);
2255 iput_mirr_err_out:
2256 #ifdef NTFS_RW
2257 	if (vol->mftmirr_ino)
2258 		iput(vol->mftmirr_ino);
2259 #endif /* NTFS_RW */
2260 	return false;
2261 }
2262 
2263 /**
2264  * ntfs_put_super - called by the vfs to unmount a volume
2265  * @sb:		vfs superblock of volume to unmount
2266  *
2267  * ntfs_put_super() is called by the VFS (from fs/super.c::do_umount()) when
2268  * the volume is being unmounted (umount system call has been invoked) and it
2269  * releases all inodes and memory belonging to the NTFS specific part of the
2270  * super block.
2271  */
2272 static void ntfs_put_super(struct super_block *sb)
2273 {
2274 	ntfs_volume *vol = NTFS_SB(sb);
2275 
2276 	ntfs_debug("Entering.");
2277 
2278 #ifdef NTFS_RW
2279 	/*
2280 	 * Commit all inodes while they are still open in case some of them
2281 	 * cause others to be dirtied.
2282 	 */
2283 	ntfs_commit_inode(vol->vol_ino);
2284 
2285 	/* NTFS 3.0+ specific. */
2286 	if (vol->major_ver >= 3) {
2287 		if (vol->usnjrnl_j_ino)
2288 			ntfs_commit_inode(vol->usnjrnl_j_ino);
2289 		if (vol->usnjrnl_max_ino)
2290 			ntfs_commit_inode(vol->usnjrnl_max_ino);
2291 		if (vol->usnjrnl_ino)
2292 			ntfs_commit_inode(vol->usnjrnl_ino);
2293 		if (vol->quota_q_ino)
2294 			ntfs_commit_inode(vol->quota_q_ino);
2295 		if (vol->quota_ino)
2296 			ntfs_commit_inode(vol->quota_ino);
2297 		if (vol->extend_ino)
2298 			ntfs_commit_inode(vol->extend_ino);
2299 		if (vol->secure_ino)
2300 			ntfs_commit_inode(vol->secure_ino);
2301 	}
2302 
2303 	ntfs_commit_inode(vol->root_ino);
2304 
2305 	down_write(&vol->lcnbmp_lock);
2306 	ntfs_commit_inode(vol->lcnbmp_ino);
2307 	up_write(&vol->lcnbmp_lock);
2308 
2309 	down_write(&vol->mftbmp_lock);
2310 	ntfs_commit_inode(vol->mftbmp_ino);
2311 	up_write(&vol->mftbmp_lock);
2312 
2313 	if (vol->logfile_ino)
2314 		ntfs_commit_inode(vol->logfile_ino);
2315 
2316 	if (vol->mftmirr_ino)
2317 		ntfs_commit_inode(vol->mftmirr_ino);
2318 	ntfs_commit_inode(vol->mft_ino);
2319 
2320 	/*
2321 	 * If a read-write mount and no volume errors have occurred, mark the
2322 	 * volume clean.  Also, re-commit all affected inodes.
2323 	 */
2324 	if (!(sb->s_flags & MS_RDONLY)) {
2325 		if (!NVolErrors(vol)) {
2326 			if (ntfs_clear_volume_flags(vol, VOLUME_IS_DIRTY))
2327 				ntfs_warning(sb, "Failed to clear dirty bit "
2328 						"in volume information "
2329 						"flags.  Run chkdsk.");
2330 			ntfs_commit_inode(vol->vol_ino);
2331 			ntfs_commit_inode(vol->root_ino);
2332 			if (vol->mftmirr_ino)
2333 				ntfs_commit_inode(vol->mftmirr_ino);
2334 			ntfs_commit_inode(vol->mft_ino);
2335 		} else {
2336 			ntfs_warning(sb, "Volume has errors.  Leaving volume "
2337 					"marked dirty.  Run chkdsk.");
2338 		}
2339 	}
2340 #endif /* NTFS_RW */
2341 
2342 	iput(vol->vol_ino);
2343 	vol->vol_ino = NULL;
2344 
2345 	/* NTFS 3.0+ specific clean up. */
2346 	if (vol->major_ver >= 3) {
2347 #ifdef NTFS_RW
2348 		if (vol->usnjrnl_j_ino) {
2349 			iput(vol->usnjrnl_j_ino);
2350 			vol->usnjrnl_j_ino = NULL;
2351 		}
2352 		if (vol->usnjrnl_max_ino) {
2353 			iput(vol->usnjrnl_max_ino);
2354 			vol->usnjrnl_max_ino = NULL;
2355 		}
2356 		if (vol->usnjrnl_ino) {
2357 			iput(vol->usnjrnl_ino);
2358 			vol->usnjrnl_ino = NULL;
2359 		}
2360 		if (vol->quota_q_ino) {
2361 			iput(vol->quota_q_ino);
2362 			vol->quota_q_ino = NULL;
2363 		}
2364 		if (vol->quota_ino) {
2365 			iput(vol->quota_ino);
2366 			vol->quota_ino = NULL;
2367 		}
2368 #endif /* NTFS_RW */
2369 		if (vol->extend_ino) {
2370 			iput(vol->extend_ino);
2371 			vol->extend_ino = NULL;
2372 		}
2373 		if (vol->secure_ino) {
2374 			iput(vol->secure_ino);
2375 			vol->secure_ino = NULL;
2376 		}
2377 	}
2378 
2379 	iput(vol->root_ino);
2380 	vol->root_ino = NULL;
2381 
2382 	down_write(&vol->lcnbmp_lock);
2383 	iput(vol->lcnbmp_ino);
2384 	vol->lcnbmp_ino = NULL;
2385 	up_write(&vol->lcnbmp_lock);
2386 
2387 	down_write(&vol->mftbmp_lock);
2388 	iput(vol->mftbmp_ino);
2389 	vol->mftbmp_ino = NULL;
2390 	up_write(&vol->mftbmp_lock);
2391 
2392 #ifdef NTFS_RW
2393 	if (vol->logfile_ino) {
2394 		iput(vol->logfile_ino);
2395 		vol->logfile_ino = NULL;
2396 	}
2397 	if (vol->mftmirr_ino) {
2398 		/* Re-commit the mft mirror and mft just in case. */
2399 		ntfs_commit_inode(vol->mftmirr_ino);
2400 		ntfs_commit_inode(vol->mft_ino);
2401 		iput(vol->mftmirr_ino);
2402 		vol->mftmirr_ino = NULL;
2403 	}
2404 	/*
2405 	 * We should have no dirty inodes left, due to
2406 	 * mft.c::ntfs_mft_writepage() cleaning all the dirty pages as
2407 	 * the underlying mft records are written out and cleaned.
2408 	 */
2409 	ntfs_commit_inode(vol->mft_ino);
2410 	write_inode_now(vol->mft_ino, 1);
2411 #endif /* NTFS_RW */
2412 
2413 	iput(vol->mft_ino);
2414 	vol->mft_ino = NULL;
2415 
2416 	/* Throw away the table of attribute definitions. */
2417 	vol->attrdef_size = 0;
2418 	if (vol->attrdef) {
2419 		ntfs_free(vol->attrdef);
2420 		vol->attrdef = NULL;
2421 	}
2422 	vol->upcase_len = 0;
2423 	/*
2424 	 * Destroy the global default upcase table if necessary.  Also decrease
2425 	 * the number of upcase users if we are a user.
2426 	 */
2427 	mutex_lock(&ntfs_lock);
2428 	if (vol->upcase == default_upcase) {
2429 		ntfs_nr_upcase_users--;
2430 		vol->upcase = NULL;
2431 	}
2432 	if (!ntfs_nr_upcase_users && default_upcase) {
2433 		ntfs_free(default_upcase);
2434 		default_upcase = NULL;
2435 	}
2436 	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
2437 		free_compression_buffers();
2438 	mutex_unlock(&ntfs_lock);
2439 	if (vol->upcase) {
2440 		ntfs_free(vol->upcase);
2441 		vol->upcase = NULL;
2442 	}
2443 
2444 	unload_nls(vol->nls_map);
2445 
2446 	sb->s_fs_info = NULL;
2447 	kfree(vol);
2448 }
2449 
2450 /**
2451  * get_nr_free_clusters - return the number of free clusters on a volume
2452  * @vol:	ntfs volume for which to obtain free cluster count
2453  *
2454  * Calculate the number of free clusters on the mounted NTFS volume @vol. We
2455  * actually calculate the number of clusters in use instead because this
2456  * allows us to not care about partial pages as these will be just zero filled
2457  * and hence not be counted as allocated clusters.
2458  *
2459  * The only particularity is that clusters beyond the end of the logical ntfs
2460  * volume will be marked as allocated to prevent errors which means we have to
2461  * discount those at the end. This is important as the cluster bitmap always
2462  * has a size in multiples of 8 bytes, i.e. up to 63 clusters could be outside
2463  * the logical volume and marked in use when they are not as they do not exist.
2464  *
2465  * If any pages cannot be read we assume all clusters in the erroring pages are
2466  * in use. This means we return an underestimate on errors which is better than
2467  * an overestimate.
2468  */
2469 static s64 get_nr_free_clusters(ntfs_volume *vol)
2470 {
2471 	s64 nr_free = vol->nr_clusters;
2472 	struct address_space *mapping = vol->lcnbmp_ino->i_mapping;
2473 	struct page *page;
2474 	pgoff_t index, max_index;
2475 
2476 	ntfs_debug("Entering.");
2477 	/* Serialize accesses to the cluster bitmap. */
2478 	down_read(&vol->lcnbmp_lock);
2479 	/*
2480 	 * Convert the number of bits into bytes rounded up, then convert into
2481 	 * multiples of PAGE_CACHE_SIZE, rounding up so that if we have one
2482 	 * full and one partial page max_index = 2.
2483 	 */
2484 	max_index = (((vol->nr_clusters + 7) >> 3) + PAGE_CACHE_SIZE - 1) >>
2485 			PAGE_CACHE_SHIFT;
2486 	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2487 	ntfs_debug("Reading $Bitmap, max_index = 0x%lx, max_size = 0x%lx.",
2488 			max_index, PAGE_CACHE_SIZE / 4);
2489 	for (index = 0; index < max_index; index++) {
2490 		unsigned long *kaddr;
2491 
2492 		/*
2493 		 * Read the page from page cache, getting it from backing store
2494 		 * if necessary, and increment the use count.
2495 		 */
2496 		page = read_mapping_page(mapping, index, NULL);
2497 		/* Ignore pages which errored synchronously. */
2498 		if (IS_ERR(page)) {
2499 			ntfs_debug("read_mapping_page() error. Skipping "
2500 					"page (index 0x%lx).", index);
2501 			nr_free -= PAGE_CACHE_SIZE * 8;
2502 			continue;
2503 		}
2504 		kaddr = kmap_atomic(page);
2505 		/*
2506 		 * Subtract the number of set bits. If this
2507 		 * is the last page and it is partial we don't really care as
2508 		 * it just means we do a little extra work but it won't affect
2509 		 * the result as all out of range bytes are set to zero by
2510 		 * ntfs_readpage().
2511 		 */
2512 		nr_free -= bitmap_weight(kaddr,
2513 					PAGE_CACHE_SIZE * BITS_PER_BYTE);
2514 		kunmap_atomic(kaddr);
2515 		page_cache_release(page);
2516 	}
2517 	ntfs_debug("Finished reading $Bitmap, last index = 0x%lx.", index - 1);
2518 	/*
2519 	 * Fixup for eventual bits outside logical ntfs volume (see function
2520 	 * description above).
2521 	 */
2522 	if (vol->nr_clusters & 63)
2523 		nr_free += 64 - (vol->nr_clusters & 63);
2524 	up_read(&vol->lcnbmp_lock);
2525 	/* If errors occurred we may well have gone below zero, fix this. */
2526 	if (nr_free < 0)
2527 		nr_free = 0;
2528 	ntfs_debug("Exiting.");
2529 	return nr_free;
2530 }
2531 
2532 /**
2533  * __get_nr_free_mft_records - return the number of free inodes on a volume
2534  * @vol:	ntfs volume for which to obtain free inode count
2535  * @nr_free:	number of mft records in filesystem
2536  * @max_index:	maximum number of pages containing set bits
2537  *
2538  * Calculate the number of free mft records (inodes) on the mounted NTFS
2539  * volume @vol. We actually calculate the number of mft records in use instead
2540  * because this allows us to not care about partial pages as these will be just
2541  * zero filled and hence not be counted as allocated mft record.
2542  *
2543  * If any pages cannot be read we assume all mft records in the erroring pages
2544  * are in use. This means we return an underestimate on errors which is better
2545  * than an overestimate.
2546  *
2547  * NOTE: Caller must hold mftbmp_lock rw_semaphore for reading or writing.
2548  */
2549 static unsigned long __get_nr_free_mft_records(ntfs_volume *vol,
2550 		s64 nr_free, const pgoff_t max_index)
2551 {
2552 	struct address_space *mapping = vol->mftbmp_ino->i_mapping;
2553 	struct page *page;
2554 	pgoff_t index;
2555 
2556 	ntfs_debug("Entering.");
2557 	/* Use multiples of 4 bytes, thus max_size is PAGE_CACHE_SIZE / 4. */
2558 	ntfs_debug("Reading $MFT/$BITMAP, max_index = 0x%lx, max_size = "
2559 			"0x%lx.", max_index, PAGE_CACHE_SIZE / 4);
2560 	for (index = 0; index < max_index; index++) {
2561 		unsigned long *kaddr;
2562 
2563 		/*
2564 		 * Read the page from page cache, getting it from backing store
2565 		 * if necessary, and increment the use count.
2566 		 */
2567 		page = read_mapping_page(mapping, index, NULL);
2568 		/* Ignore pages which errored synchronously. */
2569 		if (IS_ERR(page)) {
2570 			ntfs_debug("read_mapping_page() error. Skipping "
2571 					"page (index 0x%lx).", index);
2572 			nr_free -= PAGE_CACHE_SIZE * 8;
2573 			continue;
2574 		}
2575 		kaddr = kmap_atomic(page);
2576 		/*
2577 		 * Subtract the number of set bits. If this
2578 		 * is the last page and it is partial we don't really care as
2579 		 * it just means we do a little extra work but it won't affect
2580 		 * the result as all out of range bytes are set to zero by
2581 		 * ntfs_readpage().
2582 		 */
2583 		nr_free -= bitmap_weight(kaddr,
2584 					PAGE_CACHE_SIZE * BITS_PER_BYTE);
2585 		kunmap_atomic(kaddr);
2586 		page_cache_release(page);
2587 	}
2588 	ntfs_debug("Finished reading $MFT/$BITMAP, last index = 0x%lx.",
2589 			index - 1);
2590 	/* If errors occurred we may well have gone below zero, fix this. */
2591 	if (nr_free < 0)
2592 		nr_free = 0;
2593 	ntfs_debug("Exiting.");
2594 	return nr_free;
2595 }
2596 
2597 /**
2598  * ntfs_statfs - return information about mounted NTFS volume
2599  * @dentry:	dentry from mounted volume
2600  * @sfs:	statfs structure in which to return the information
2601  *
2602  * Return information about the mounted NTFS volume @dentry in the statfs structure
2603  * pointed to by @sfs (this is initialized with zeros before ntfs_statfs is
2604  * called). We interpret the values to be correct of the moment in time at
2605  * which we are called. Most values are variable otherwise and this isn't just
2606  * the free values but the totals as well. For example we can increase the
2607  * total number of file nodes if we run out and we can keep doing this until
2608  * there is no more space on the volume left at all.
2609  *
2610  * Called from vfs_statfs which is used to handle the statfs, fstatfs, and
2611  * ustat system calls.
2612  *
2613  * Return 0 on success or -errno on error.
2614  */
2615 static int ntfs_statfs(struct dentry *dentry, struct kstatfs *sfs)
2616 {
2617 	struct super_block *sb = dentry->d_sb;
2618 	s64 size;
2619 	ntfs_volume *vol = NTFS_SB(sb);
2620 	ntfs_inode *mft_ni = NTFS_I(vol->mft_ino);
2621 	pgoff_t max_index;
2622 	unsigned long flags;
2623 
2624 	ntfs_debug("Entering.");
2625 	/* Type of filesystem. */
2626 	sfs->f_type   = NTFS_SB_MAGIC;
2627 	/* Optimal transfer block size. */
2628 	sfs->f_bsize  = PAGE_CACHE_SIZE;
2629 	/*
2630 	 * Total data blocks in filesystem in units of f_bsize and since
2631 	 * inodes are also stored in data blocs ($MFT is a file) this is just
2632 	 * the total clusters.
2633 	 */
2634 	sfs->f_blocks = vol->nr_clusters << vol->cluster_size_bits >>
2635 				PAGE_CACHE_SHIFT;
2636 	/* Free data blocks in filesystem in units of f_bsize. */
2637 	size	      = get_nr_free_clusters(vol) << vol->cluster_size_bits >>
2638 				PAGE_CACHE_SHIFT;
2639 	if (size < 0LL)
2640 		size = 0LL;
2641 	/* Free blocks avail to non-superuser, same as above on NTFS. */
2642 	sfs->f_bavail = sfs->f_bfree = size;
2643 	/* Serialize accesses to the inode bitmap. */
2644 	down_read(&vol->mftbmp_lock);
2645 	read_lock_irqsave(&mft_ni->size_lock, flags);
2646 	size = i_size_read(vol->mft_ino) >> vol->mft_record_size_bits;
2647 	/*
2648 	 * Convert the maximum number of set bits into bytes rounded up, then
2649 	 * convert into multiples of PAGE_CACHE_SIZE, rounding up so that if we
2650 	 * have one full and one partial page max_index = 2.
2651 	 */
2652 	max_index = ((((mft_ni->initialized_size >> vol->mft_record_size_bits)
2653 			+ 7) >> 3) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
2654 	read_unlock_irqrestore(&mft_ni->size_lock, flags);
2655 	/* Number of inodes in filesystem (at this point in time). */
2656 	sfs->f_files = size;
2657 	/* Free inodes in fs (based on current total count). */
2658 	sfs->f_ffree = __get_nr_free_mft_records(vol, size, max_index);
2659 	up_read(&vol->mftbmp_lock);
2660 	/*
2661 	 * File system id. This is extremely *nix flavour dependent and even
2662 	 * within Linux itself all fs do their own thing. I interpret this to
2663 	 * mean a unique id associated with the mounted fs and not the id
2664 	 * associated with the filesystem driver, the latter is already given
2665 	 * by the filesystem type in sfs->f_type. Thus we use the 64-bit
2666 	 * volume serial number splitting it into two 32-bit parts. We enter
2667 	 * the least significant 32-bits in f_fsid[0] and the most significant
2668 	 * 32-bits in f_fsid[1].
2669 	 */
2670 	sfs->f_fsid.val[0] = vol->serial_no & 0xffffffff;
2671 	sfs->f_fsid.val[1] = (vol->serial_no >> 32) & 0xffffffff;
2672 	/* Maximum length of filenames. */
2673 	sfs->f_namelen	   = NTFS_MAX_NAME_LEN;
2674 	return 0;
2675 }
2676 
2677 #ifdef NTFS_RW
2678 static int ntfs_write_inode(struct inode *vi, struct writeback_control *wbc)
2679 {
2680 	return __ntfs_write_inode(vi, wbc->sync_mode == WB_SYNC_ALL);
2681 }
2682 #endif
2683 
2684 /**
2685  * The complete super operations.
2686  */
2687 static const struct super_operations ntfs_sops = {
2688 	.alloc_inode	= ntfs_alloc_big_inode,	  /* VFS: Allocate new inode. */
2689 	.destroy_inode	= ntfs_destroy_big_inode, /* VFS: Deallocate inode. */
2690 #ifdef NTFS_RW
2691 	.write_inode	= ntfs_write_inode,	/* VFS: Write dirty inode to
2692 						   disk. */
2693 #endif /* NTFS_RW */
2694 	.put_super	= ntfs_put_super,	/* Syscall: umount. */
2695 	.statfs		= ntfs_statfs,		/* Syscall: statfs */
2696 	.remount_fs	= ntfs_remount,		/* Syscall: mount -o remount. */
2697 	.evict_inode	= ntfs_evict_big_inode,	/* VFS: Called when an inode is
2698 						   removed from memory. */
2699 	.show_options	= ntfs_show_options,	/* Show mount options in
2700 						   proc. */
2701 };
2702 
2703 /**
2704  * ntfs_fill_super - mount an ntfs filesystem
2705  * @sb:		super block of ntfs filesystem to mount
2706  * @opt:	string containing the mount options
2707  * @silent:	silence error output
2708  *
2709  * ntfs_fill_super() is called by the VFS to mount the device described by @sb
2710  * with the mount otions in @data with the NTFS filesystem.
2711  *
2712  * If @silent is true, remain silent even if errors are detected. This is used
2713  * during bootup, when the kernel tries to mount the root filesystem with all
2714  * registered filesystems one after the other until one succeeds. This implies
2715  * that all filesystems except the correct one will quite correctly and
2716  * expectedly return an error, but nobody wants to see error messages when in
2717  * fact this is what is supposed to happen.
2718  *
2719  * NOTE: @sb->s_flags contains the mount options flags.
2720  */
2721 static int ntfs_fill_super(struct super_block *sb, void *opt, const int silent)
2722 {
2723 	ntfs_volume *vol;
2724 	struct buffer_head *bh;
2725 	struct inode *tmp_ino;
2726 	int blocksize, result;
2727 
2728 	/*
2729 	 * We do a pretty difficult piece of bootstrap by reading the
2730 	 * MFT (and other metadata) from disk into memory. We'll only
2731 	 * release this metadata during umount, so the locking patterns
2732 	 * observed during bootstrap do not count. So turn off the
2733 	 * observation of locking patterns (strictly for this context
2734 	 * only) while mounting NTFS. [The validator is still active
2735 	 * otherwise, even for this context: it will for example record
2736 	 * lock class registrations.]
2737 	 */
2738 	lockdep_off();
2739 	ntfs_debug("Entering.");
2740 #ifndef NTFS_RW
2741 	sb->s_flags |= MS_RDONLY;
2742 #endif /* ! NTFS_RW */
2743 	/* Allocate a new ntfs_volume and place it in sb->s_fs_info. */
2744 	sb->s_fs_info = kmalloc(sizeof(ntfs_volume), GFP_NOFS);
2745 	vol = NTFS_SB(sb);
2746 	if (!vol) {
2747 		if (!silent)
2748 			ntfs_error(sb, "Allocation of NTFS volume structure "
2749 					"failed. Aborting mount...");
2750 		lockdep_on();
2751 		return -ENOMEM;
2752 	}
2753 	/* Initialize ntfs_volume structure. */
2754 	*vol = (ntfs_volume) {
2755 		.sb = sb,
2756 		/*
2757 		 * Default is group and other don't have any access to files or
2758 		 * directories while owner has full access. Further, files by
2759 		 * default are not executable but directories are of course
2760 		 * browseable.
2761 		 */
2762 		.fmask = 0177,
2763 		.dmask = 0077,
2764 	};
2765 	init_rwsem(&vol->mftbmp_lock);
2766 	init_rwsem(&vol->lcnbmp_lock);
2767 
2768 	/* By default, enable sparse support. */
2769 	NVolSetSparseEnabled(vol);
2770 
2771 	/* Important to get the mount options dealt with now. */
2772 	if (!parse_options(vol, (char*)opt))
2773 		goto err_out_now;
2774 
2775 	/* We support sector sizes up to the PAGE_CACHE_SIZE. */
2776 	if (bdev_logical_block_size(sb->s_bdev) > PAGE_CACHE_SIZE) {
2777 		if (!silent)
2778 			ntfs_error(sb, "Device has unsupported sector size "
2779 					"(%i).  The maximum supported sector "
2780 					"size on this architecture is %lu "
2781 					"bytes.",
2782 					bdev_logical_block_size(sb->s_bdev),
2783 					PAGE_CACHE_SIZE);
2784 		goto err_out_now;
2785 	}
2786 	/*
2787 	 * Setup the device access block size to NTFS_BLOCK_SIZE or the hard
2788 	 * sector size, whichever is bigger.
2789 	 */
2790 	blocksize = sb_min_blocksize(sb, NTFS_BLOCK_SIZE);
2791 	if (blocksize < NTFS_BLOCK_SIZE) {
2792 		if (!silent)
2793 			ntfs_error(sb, "Unable to set device block size.");
2794 		goto err_out_now;
2795 	}
2796 	BUG_ON(blocksize != sb->s_blocksize);
2797 	ntfs_debug("Set device block size to %i bytes (block size bits %i).",
2798 			blocksize, sb->s_blocksize_bits);
2799 	/* Determine the size of the device in units of block_size bytes. */
2800 	if (!i_size_read(sb->s_bdev->bd_inode)) {
2801 		if (!silent)
2802 			ntfs_error(sb, "Unable to determine device size.");
2803 		goto err_out_now;
2804 	}
2805 	vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2806 			sb->s_blocksize_bits;
2807 	/* Read the boot sector and return unlocked buffer head to it. */
2808 	if (!(bh = read_ntfs_boot_sector(sb, silent))) {
2809 		if (!silent)
2810 			ntfs_error(sb, "Not an NTFS volume.");
2811 		goto err_out_now;
2812 	}
2813 	/*
2814 	 * Extract the data from the boot sector and setup the ntfs volume
2815 	 * using it.
2816 	 */
2817 	result = parse_ntfs_boot_sector(vol, (NTFS_BOOT_SECTOR*)bh->b_data);
2818 	brelse(bh);
2819 	if (!result) {
2820 		if (!silent)
2821 			ntfs_error(sb, "Unsupported NTFS filesystem.");
2822 		goto err_out_now;
2823 	}
2824 	/*
2825 	 * If the boot sector indicates a sector size bigger than the current
2826 	 * device block size, switch the device block size to the sector size.
2827 	 * TODO: It may be possible to support this case even when the set
2828 	 * below fails, we would just be breaking up the i/o for each sector
2829 	 * into multiple blocks for i/o purposes but otherwise it should just
2830 	 * work.  However it is safer to leave disabled until someone hits this
2831 	 * error message and then we can get them to try it without the setting
2832 	 * so we know for sure that it works.
2833 	 */
2834 	if (vol->sector_size > blocksize) {
2835 		blocksize = sb_set_blocksize(sb, vol->sector_size);
2836 		if (blocksize != vol->sector_size) {
2837 			if (!silent)
2838 				ntfs_error(sb, "Unable to set device block "
2839 						"size to sector size (%i).",
2840 						vol->sector_size);
2841 			goto err_out_now;
2842 		}
2843 		BUG_ON(blocksize != sb->s_blocksize);
2844 		vol->nr_blocks = i_size_read(sb->s_bdev->bd_inode) >>
2845 				sb->s_blocksize_bits;
2846 		ntfs_debug("Changed device block size to %i bytes (block size "
2847 				"bits %i) to match volume sector size.",
2848 				blocksize, sb->s_blocksize_bits);
2849 	}
2850 	/* Initialize the cluster and mft allocators. */
2851 	ntfs_setup_allocators(vol);
2852 	/* Setup remaining fields in the super block. */
2853 	sb->s_magic = NTFS_SB_MAGIC;
2854 	/*
2855 	 * Ntfs allows 63 bits for the file size, i.e. correct would be:
2856 	 *	sb->s_maxbytes = ~0ULL >> 1;
2857 	 * But the kernel uses a long as the page cache page index which on
2858 	 * 32-bit architectures is only 32-bits. MAX_LFS_FILESIZE is kernel
2859 	 * defined to the maximum the page cache page index can cope with
2860 	 * without overflowing the index or to 2^63 - 1, whichever is smaller.
2861 	 */
2862 	sb->s_maxbytes = MAX_LFS_FILESIZE;
2863 	/* Ntfs measures time in 100ns intervals. */
2864 	sb->s_time_gran = 100;
2865 	/*
2866 	 * Now load the metadata required for the page cache and our address
2867 	 * space operations to function. We do this by setting up a specialised
2868 	 * read_inode method and then just calling the normal iget() to obtain
2869 	 * the inode for $MFT which is sufficient to allow our normal inode
2870 	 * operations and associated address space operations to function.
2871 	 */
2872 	sb->s_op = &ntfs_sops;
2873 	tmp_ino = new_inode(sb);
2874 	if (!tmp_ino) {
2875 		if (!silent)
2876 			ntfs_error(sb, "Failed to load essential metadata.");
2877 		goto err_out_now;
2878 	}
2879 	tmp_ino->i_ino = FILE_MFT;
2880 	insert_inode_hash(tmp_ino);
2881 	if (ntfs_read_inode_mount(tmp_ino) < 0) {
2882 		if (!silent)
2883 			ntfs_error(sb, "Failed to load essential metadata.");
2884 		goto iput_tmp_ino_err_out_now;
2885 	}
2886 	mutex_lock(&ntfs_lock);
2887 	/*
2888 	 * The current mount is a compression user if the cluster size is
2889 	 * less than or equal 4kiB.
2890 	 */
2891 	if (vol->cluster_size <= 4096 && !ntfs_nr_compression_users++) {
2892 		result = allocate_compression_buffers();
2893 		if (result) {
2894 			ntfs_error(NULL, "Failed to allocate buffers "
2895 					"for compression engine.");
2896 			ntfs_nr_compression_users--;
2897 			mutex_unlock(&ntfs_lock);
2898 			goto iput_tmp_ino_err_out_now;
2899 		}
2900 	}
2901 	/*
2902 	 * Generate the global default upcase table if necessary.  Also
2903 	 * temporarily increment the number of upcase users to avoid race
2904 	 * conditions with concurrent (u)mounts.
2905 	 */
2906 	if (!default_upcase)
2907 		default_upcase = generate_default_upcase();
2908 	ntfs_nr_upcase_users++;
2909 	mutex_unlock(&ntfs_lock);
2910 	/*
2911 	 * From now on, ignore @silent parameter. If we fail below this line,
2912 	 * it will be due to a corrupt fs or a system error, so we report it.
2913 	 */
2914 	/*
2915 	 * Open the system files with normal access functions and complete
2916 	 * setting up the ntfs super block.
2917 	 */
2918 	if (!load_system_files(vol)) {
2919 		ntfs_error(sb, "Failed to load system files.");
2920 		goto unl_upcase_iput_tmp_ino_err_out_now;
2921 	}
2922 
2923 	/* We grab a reference, simulating an ntfs_iget(). */
2924 	ihold(vol->root_ino);
2925 	if ((sb->s_root = d_make_root(vol->root_ino))) {
2926 		ntfs_debug("Exiting, status successful.");
2927 		/* Release the default upcase if it has no users. */
2928 		mutex_lock(&ntfs_lock);
2929 		if (!--ntfs_nr_upcase_users && default_upcase) {
2930 			ntfs_free(default_upcase);
2931 			default_upcase = NULL;
2932 		}
2933 		mutex_unlock(&ntfs_lock);
2934 		sb->s_export_op = &ntfs_export_ops;
2935 		lockdep_on();
2936 		return 0;
2937 	}
2938 	ntfs_error(sb, "Failed to allocate root directory.");
2939 	/* Clean up after the successful load_system_files() call from above. */
2940 	// TODO: Use ntfs_put_super() instead of repeating all this code...
2941 	// FIXME: Should mark the volume clean as the error is most likely
2942 	// 	  -ENOMEM.
2943 	iput(vol->vol_ino);
2944 	vol->vol_ino = NULL;
2945 	/* NTFS 3.0+ specific clean up. */
2946 	if (vol->major_ver >= 3) {
2947 #ifdef NTFS_RW
2948 		if (vol->usnjrnl_j_ino) {
2949 			iput(vol->usnjrnl_j_ino);
2950 			vol->usnjrnl_j_ino = NULL;
2951 		}
2952 		if (vol->usnjrnl_max_ino) {
2953 			iput(vol->usnjrnl_max_ino);
2954 			vol->usnjrnl_max_ino = NULL;
2955 		}
2956 		if (vol->usnjrnl_ino) {
2957 			iput(vol->usnjrnl_ino);
2958 			vol->usnjrnl_ino = NULL;
2959 		}
2960 		if (vol->quota_q_ino) {
2961 			iput(vol->quota_q_ino);
2962 			vol->quota_q_ino = NULL;
2963 		}
2964 		if (vol->quota_ino) {
2965 			iput(vol->quota_ino);
2966 			vol->quota_ino = NULL;
2967 		}
2968 #endif /* NTFS_RW */
2969 		if (vol->extend_ino) {
2970 			iput(vol->extend_ino);
2971 			vol->extend_ino = NULL;
2972 		}
2973 		if (vol->secure_ino) {
2974 			iput(vol->secure_ino);
2975 			vol->secure_ino = NULL;
2976 		}
2977 	}
2978 	iput(vol->root_ino);
2979 	vol->root_ino = NULL;
2980 	iput(vol->lcnbmp_ino);
2981 	vol->lcnbmp_ino = NULL;
2982 	iput(vol->mftbmp_ino);
2983 	vol->mftbmp_ino = NULL;
2984 #ifdef NTFS_RW
2985 	if (vol->logfile_ino) {
2986 		iput(vol->logfile_ino);
2987 		vol->logfile_ino = NULL;
2988 	}
2989 	if (vol->mftmirr_ino) {
2990 		iput(vol->mftmirr_ino);
2991 		vol->mftmirr_ino = NULL;
2992 	}
2993 #endif /* NTFS_RW */
2994 	/* Throw away the table of attribute definitions. */
2995 	vol->attrdef_size = 0;
2996 	if (vol->attrdef) {
2997 		ntfs_free(vol->attrdef);
2998 		vol->attrdef = NULL;
2999 	}
3000 	vol->upcase_len = 0;
3001 	mutex_lock(&ntfs_lock);
3002 	if (vol->upcase == default_upcase) {
3003 		ntfs_nr_upcase_users--;
3004 		vol->upcase = NULL;
3005 	}
3006 	mutex_unlock(&ntfs_lock);
3007 	if (vol->upcase) {
3008 		ntfs_free(vol->upcase);
3009 		vol->upcase = NULL;
3010 	}
3011 	if (vol->nls_map) {
3012 		unload_nls(vol->nls_map);
3013 		vol->nls_map = NULL;
3014 	}
3015 	/* Error exit code path. */
3016 unl_upcase_iput_tmp_ino_err_out_now:
3017 	/*
3018 	 * Decrease the number of upcase users and destroy the global default
3019 	 * upcase table if necessary.
3020 	 */
3021 	mutex_lock(&ntfs_lock);
3022 	if (!--ntfs_nr_upcase_users && default_upcase) {
3023 		ntfs_free(default_upcase);
3024 		default_upcase = NULL;
3025 	}
3026 	if (vol->cluster_size <= 4096 && !--ntfs_nr_compression_users)
3027 		free_compression_buffers();
3028 	mutex_unlock(&ntfs_lock);
3029 iput_tmp_ino_err_out_now:
3030 	iput(tmp_ino);
3031 	if (vol->mft_ino && vol->mft_ino != tmp_ino)
3032 		iput(vol->mft_ino);
3033 	vol->mft_ino = NULL;
3034 	/* Errors at this stage are irrelevant. */
3035 err_out_now:
3036 	sb->s_fs_info = NULL;
3037 	kfree(vol);
3038 	ntfs_debug("Failed, returning -EINVAL.");
3039 	lockdep_on();
3040 	return -EINVAL;
3041 }
3042 
3043 /*
3044  * This is a slab cache to optimize allocations and deallocations of Unicode
3045  * strings of the maximum length allowed by NTFS, which is NTFS_MAX_NAME_LEN
3046  * (255) Unicode characters + a terminating NULL Unicode character.
3047  */
3048 struct kmem_cache *ntfs_name_cache;
3049 
3050 /* Slab caches for efficient allocation/deallocation of inodes. */
3051 struct kmem_cache *ntfs_inode_cache;
3052 struct kmem_cache *ntfs_big_inode_cache;
3053 
3054 /* Init once constructor for the inode slab cache. */
3055 static void ntfs_big_inode_init_once(void *foo)
3056 {
3057 	ntfs_inode *ni = (ntfs_inode *)foo;
3058 
3059 	inode_init_once(VFS_I(ni));
3060 }
3061 
3062 /*
3063  * Slab caches to optimize allocations and deallocations of attribute search
3064  * contexts and index contexts, respectively.
3065  */
3066 struct kmem_cache *ntfs_attr_ctx_cache;
3067 struct kmem_cache *ntfs_index_ctx_cache;
3068 
3069 /* Driver wide mutex. */
3070 DEFINE_MUTEX(ntfs_lock);
3071 
3072 static struct dentry *ntfs_mount(struct file_system_type *fs_type,
3073 	int flags, const char *dev_name, void *data)
3074 {
3075 	return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
3076 }
3077 
3078 static struct file_system_type ntfs_fs_type = {
3079 	.owner		= THIS_MODULE,
3080 	.name		= "ntfs",
3081 	.mount		= ntfs_mount,
3082 	.kill_sb	= kill_block_super,
3083 	.fs_flags	= FS_REQUIRES_DEV,
3084 };
3085 MODULE_ALIAS_FS("ntfs");
3086 
3087 /* Stable names for the slab caches. */
3088 static const char ntfs_index_ctx_cache_name[] = "ntfs_index_ctx_cache";
3089 static const char ntfs_attr_ctx_cache_name[] = "ntfs_attr_ctx_cache";
3090 static const char ntfs_name_cache_name[] = "ntfs_name_cache";
3091 static const char ntfs_inode_cache_name[] = "ntfs_inode_cache";
3092 static const char ntfs_big_inode_cache_name[] = "ntfs_big_inode_cache";
3093 
3094 static int __init init_ntfs_fs(void)
3095 {
3096 	int err = 0;
3097 
3098 	/* This may be ugly but it results in pretty output so who cares. (-8 */
3099 	pr_info("driver " NTFS_VERSION " [Flags: R/"
3100 #ifdef NTFS_RW
3101 			"W"
3102 #else
3103 			"O"
3104 #endif
3105 #ifdef DEBUG
3106 			" DEBUG"
3107 #endif
3108 #ifdef MODULE
3109 			" MODULE"
3110 #endif
3111 			"].\n");
3112 
3113 	ntfs_debug("Debug messages are enabled.");
3114 
3115 	ntfs_index_ctx_cache = kmem_cache_create(ntfs_index_ctx_cache_name,
3116 			sizeof(ntfs_index_context), 0 /* offset */,
3117 			SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3118 	if (!ntfs_index_ctx_cache) {
3119 		pr_crit("Failed to create %s!\n", ntfs_index_ctx_cache_name);
3120 		goto ictx_err_out;
3121 	}
3122 	ntfs_attr_ctx_cache = kmem_cache_create(ntfs_attr_ctx_cache_name,
3123 			sizeof(ntfs_attr_search_ctx), 0 /* offset */,
3124 			SLAB_HWCACHE_ALIGN, NULL /* ctor */);
3125 	if (!ntfs_attr_ctx_cache) {
3126 		pr_crit("NTFS: Failed to create %s!\n",
3127 			ntfs_attr_ctx_cache_name);
3128 		goto actx_err_out;
3129 	}
3130 
3131 	ntfs_name_cache = kmem_cache_create(ntfs_name_cache_name,
3132 			(NTFS_MAX_NAME_LEN+1) * sizeof(ntfschar), 0,
3133 			SLAB_HWCACHE_ALIGN, NULL);
3134 	if (!ntfs_name_cache) {
3135 		pr_crit("Failed to create %s!\n", ntfs_name_cache_name);
3136 		goto name_err_out;
3137 	}
3138 
3139 	ntfs_inode_cache = kmem_cache_create(ntfs_inode_cache_name,
3140 			sizeof(ntfs_inode), 0,
3141 			SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD, NULL);
3142 	if (!ntfs_inode_cache) {
3143 		pr_crit("Failed to create %s!\n", ntfs_inode_cache_name);
3144 		goto inode_err_out;
3145 	}
3146 
3147 	ntfs_big_inode_cache = kmem_cache_create(ntfs_big_inode_cache_name,
3148 			sizeof(big_ntfs_inode), 0,
3149 			SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
3150 			ntfs_big_inode_init_once);
3151 	if (!ntfs_big_inode_cache) {
3152 		pr_crit("Failed to create %s!\n", ntfs_big_inode_cache_name);
3153 		goto big_inode_err_out;
3154 	}
3155 
3156 	/* Register the ntfs sysctls. */
3157 	err = ntfs_sysctl(1);
3158 	if (err) {
3159 		pr_crit("Failed to register NTFS sysctls!\n");
3160 		goto sysctl_err_out;
3161 	}
3162 
3163 	err = register_filesystem(&ntfs_fs_type);
3164 	if (!err) {
3165 		ntfs_debug("NTFS driver registered successfully.");
3166 		return 0; /* Success! */
3167 	}
3168 	pr_crit("Failed to register NTFS filesystem driver!\n");
3169 
3170 	/* Unregister the ntfs sysctls. */
3171 	ntfs_sysctl(0);
3172 sysctl_err_out:
3173 	kmem_cache_destroy(ntfs_big_inode_cache);
3174 big_inode_err_out:
3175 	kmem_cache_destroy(ntfs_inode_cache);
3176 inode_err_out:
3177 	kmem_cache_destroy(ntfs_name_cache);
3178 name_err_out:
3179 	kmem_cache_destroy(ntfs_attr_ctx_cache);
3180 actx_err_out:
3181 	kmem_cache_destroy(ntfs_index_ctx_cache);
3182 ictx_err_out:
3183 	if (!err) {
3184 		pr_crit("Aborting NTFS filesystem driver registration...\n");
3185 		err = -ENOMEM;
3186 	}
3187 	return err;
3188 }
3189 
3190 static void __exit exit_ntfs_fs(void)
3191 {
3192 	ntfs_debug("Unregistering NTFS driver.");
3193 
3194 	unregister_filesystem(&ntfs_fs_type);
3195 
3196 	/*
3197 	 * Make sure all delayed rcu free inodes are flushed before we
3198 	 * destroy cache.
3199 	 */
3200 	rcu_barrier();
3201 	kmem_cache_destroy(ntfs_big_inode_cache);
3202 	kmem_cache_destroy(ntfs_inode_cache);
3203 	kmem_cache_destroy(ntfs_name_cache);
3204 	kmem_cache_destroy(ntfs_attr_ctx_cache);
3205 	kmem_cache_destroy(ntfs_index_ctx_cache);
3206 	/* Unregister the ntfs sysctls. */
3207 	ntfs_sysctl(0);
3208 }
3209 
3210 MODULE_AUTHOR("Anton Altaparmakov <anton@tuxera.com>");
3211 MODULE_DESCRIPTION("NTFS 1.2/3.x driver - Copyright (c) 2001-2011 Anton Altaparmakov and Tuxera Inc.");
3212 MODULE_VERSION(NTFS_VERSION);
3213 MODULE_LICENSE("GPL");
3214 #ifdef DEBUG
3215 module_param(debug_msgs, bint, 0);
3216 MODULE_PARM_DESC(debug_msgs, "Enable debug messages.");
3217 #endif
3218 
3219 module_init(init_ntfs_fs)
3220 module_exit(exit_ntfs_fs)
3221