xfs_super.c 47.7 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
3
 * All Rights Reserved.
L
Linus Torvalds 已提交
4
 *
5 6
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
L
Linus Torvalds 已提交
7 8
 * published by the Free Software Foundation.
 *
9 10 11 12
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
L
Linus Torvalds 已提交
13
 *
14 15 16
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
L
Linus Torvalds 已提交
17
 */
C
Christoph Hellwig 已提交
18

L
Linus Torvalds 已提交
19
#include "xfs.h"
20
#include "xfs_shared.h"
21
#include "xfs_format.h"
22 23
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
24
#include "xfs_inum.h"
L
Linus Torvalds 已提交
25
#include "xfs_sb.h"
26
#include "xfs_ag.h"
L
Linus Torvalds 已提交
27
#include "xfs_mount.h"
28
#include "xfs_da_format.h"
L
Linus Torvalds 已提交
29
#include "xfs_inode.h"
30
#include "xfs_btree.h"
L
Linus Torvalds 已提交
31
#include "xfs_bmap.h"
32
#include "xfs_alloc.h"
L
Linus Torvalds 已提交
33
#include "xfs_error.h"
C
Christoph Hellwig 已提交
34
#include "xfs_fsops.h"
35
#include "xfs_trans.h"
L
Linus Torvalds 已提交
36
#include "xfs_buf_item.h"
37
#include "xfs_log.h"
38
#include "xfs_log_priv.h"
39
#include "xfs_da_btree.h"
40
#include "xfs_dir2.h"
41 42 43
#include "xfs_extfree_item.h"
#include "xfs_mru_cache.h"
#include "xfs_inode_item.h"
44
#include "xfs_icache.h"
C
Christoph Hellwig 已提交
45
#include "xfs_trace.h"
D
Dave Chinner 已提交
46
#include "xfs_icreate_item.h"
47 48 49
#include "xfs_dinode.h"
#include "xfs_filestream.h"
#include "xfs_quota.h"
L
Linus Torvalds 已提交
50 51 52

#include <linux/namei.h>
#include <linux/init.h>
53
#include <linux/slab.h>
L
Linus Torvalds 已提交
54
#include <linux/mount.h>
55
#include <linux/mempool.h>
L
Linus Torvalds 已提交
56
#include <linux/writeback.h>
57
#include <linux/kthread.h>
58
#include <linux/freezer.h>
59
#include <linux/parser.h>
L
Linus Torvalds 已提交
60

61
static const struct super_operations xfs_super_operations;
62
static kmem_zone_t *xfs_ioend_zone;
63
mempool_t *xfs_ioend_pool;
L
Linus Torvalds 已提交
64

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
#define MNTOPT_LOGBUFS	"logbufs"	/* number of XFS log buffers */
#define MNTOPT_LOGBSIZE	"logbsize"	/* size of XFS log buffers */
#define MNTOPT_LOGDEV	"logdev"	/* log device */
#define MNTOPT_RTDEV	"rtdev"		/* realtime I/O device */
#define MNTOPT_BIOSIZE	"biosize"	/* log2 of preferred buffered io size */
#define MNTOPT_WSYNC	"wsync"		/* safe-mode nfs compatible mount */
#define MNTOPT_NOALIGN	"noalign"	/* turn off stripe alignment */
#define MNTOPT_SWALLOC	"swalloc"	/* turn on stripe width allocation */
#define MNTOPT_SUNIT	"sunit"		/* data volume stripe unit */
#define MNTOPT_SWIDTH	"swidth"	/* data volume stripe width */
#define MNTOPT_NOUUID	"nouuid"	/* ignore filesystem UUID */
#define MNTOPT_MTPT	"mtpt"		/* filesystem mount point */
#define MNTOPT_GRPID	"grpid"		/* group-ID from parent directory */
#define MNTOPT_NOGRPID	"nogrpid"	/* group-ID from current process */
#define MNTOPT_BSDGROUPS    "bsdgroups"    /* group-ID from parent directory */
#define MNTOPT_SYSVGROUPS   "sysvgroups"   /* group-ID from current process */
#define MNTOPT_ALLOCSIZE    "allocsize"    /* preferred allocation size */
#define MNTOPT_NORECOVERY   "norecovery"   /* don't run XFS recovery */
#define MNTOPT_BARRIER	"barrier"	/* use writer barriers for log write and
					 * unwritten extent conversion */
#define MNTOPT_NOBARRIER "nobarrier"	/* .. disable */
#define MNTOPT_64BITINODE   "inode64"	/* inodes can be allocated anywhere */
87 88
#define MNTOPT_32BITINODE   "inode32"	/* inode allocation limited to
					 * XFS_MAXINUMBER_32 */
89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
#define MNTOPT_IKEEP	"ikeep"		/* do not free empty inode clusters */
#define MNTOPT_NOIKEEP	"noikeep"	/* free empty inode clusters */
#define MNTOPT_LARGEIO	   "largeio"	/* report large I/O sizes in stat() */
#define MNTOPT_NOLARGEIO   "nolargeio"	/* do not report large I/O sizes
					 * in stat(). */
#define MNTOPT_ATTR2	"attr2"		/* do use attr2 attribute format */
#define MNTOPT_NOATTR2	"noattr2"	/* do not use attr2 attribute format */
#define MNTOPT_FILESTREAM  "filestreams" /* use filestreams allocator */
#define MNTOPT_QUOTA	"quota"		/* disk quotas (user) */
#define MNTOPT_NOQUOTA	"noquota"	/* no quotas */
#define MNTOPT_USRQUOTA	"usrquota"	/* user quota enabled */
#define MNTOPT_GRPQUOTA	"grpquota"	/* group quota enabled */
#define MNTOPT_PRJQUOTA	"prjquota"	/* project quota enabled */
#define MNTOPT_UQUOTA	"uquota"	/* user quota (IRIX variant) */
#define MNTOPT_GQUOTA	"gquota"	/* group quota (IRIX variant) */
#define MNTOPT_PQUOTA	"pquota"	/* project quota (IRIX variant) */
#define MNTOPT_UQUOTANOENF "uqnoenforce"/* user quota limit enforcement */
#define MNTOPT_GQUOTANOENF "gqnoenforce"/* group quota limit enforcement */
#define MNTOPT_PQUOTANOENF "pqnoenforce"/* project quota limit enforcement */
#define MNTOPT_QUOTANOENF  "qnoenforce"	/* same as uqnoenforce */
109 110 111 112
#define MNTOPT_DELAYLOG    "delaylog"	/* Delayed logging enabled */
#define MNTOPT_NODELAYLOG  "nodelaylog"	/* Delayed logging disabled */
#define MNTOPT_DISCARD	   "discard"	/* Discard unused blocks */
#define MNTOPT_NODISCARD   "nodiscard"	/* Do not discard unused blocks */
113

114 115 116 117 118 119 120
/*
 * Table driven mount option parser.
 *
 * Currently only used for remount, but it will be used for mount
 * in the future, too.
 */
enum {
121 122 123 124 125
	Opt_barrier,
	Opt_nobarrier,
	Opt_inode64,
	Opt_inode32,
	Opt_err
126 127
};

128
static const match_table_t tokens = {
129 130
	{Opt_barrier, "barrier"},
	{Opt_nobarrier, "nobarrier"},
131
	{Opt_inode64, "inode64"},
132
	{Opt_inode32, "inode32"},
133 134 135 136
	{Opt_err, NULL}
};


137
STATIC unsigned long
138
suffix_kstrtoint(char *s, unsigned int base, int *res)
139
{
140
	int	last, shift_left_factor = 0, _res;
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
	char	*value = s;

	last = strlen(value) - 1;
	if (value[last] == 'K' || value[last] == 'k') {
		shift_left_factor = 10;
		value[last] = '\0';
	}
	if (value[last] == 'M' || value[last] == 'm') {
		shift_left_factor = 20;
		value[last] = '\0';
	}
	if (value[last] == 'G' || value[last] == 'g') {
		shift_left_factor = 30;
		value[last] = '\0';
	}

157 158 159 160
	if (kstrtoint(s, base, &_res))
		return -EINVAL;
	*res = _res << shift_left_factor;
	return 0;
161 162
}

163 164 165 166 167 168 169
/*
 * This function fills in xfs_mount_t fields based on mount args.
 * Note: the superblock has _not_ yet been read in.
 *
 * Note that this function leaks the various device name allocations on
 * failure.  The caller takes care of them.
 */
170 171 172
STATIC int
xfs_parseargs(
	struct xfs_mount	*mp,
C
Christoph Hellwig 已提交
173
	char			*options)
174
{
175
	struct super_block	*sb = mp->m_super;
176
	char			*this_char, *value;
177 178 179
	int			dsunit = 0;
	int			dswidth = 0;
	int			iosize = 0;
180
	__uint8_t		iosizelog = 0;
181

182 183 184 185 186 187 188 189 190
	/*
	 * set up the mount name first so all the errors will refer to the
	 * correct device.
	 */
	mp->m_fsname = kstrndup(sb->s_id, MAXNAMELEN, GFP_KERNEL);
	if (!mp->m_fsname)
		return ENOMEM;
	mp->m_fsname_len = strlen(mp->m_fsname) + 1;

191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
	/*
	 * Copy binary VFS mount flags we are interested in.
	 */
	if (sb->s_flags & MS_RDONLY)
		mp->m_flags |= XFS_MOUNT_RDONLY;
	if (sb->s_flags & MS_DIRSYNC)
		mp->m_flags |= XFS_MOUNT_DIRSYNC;
	if (sb->s_flags & MS_SYNCHRONOUS)
		mp->m_flags |= XFS_MOUNT_WSYNC;

	/*
	 * Set some default flags that could be cleared by the mount option
	 * parsing.
	 */
	mp->m_flags |= XFS_MOUNT_BARRIER;
	mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
207
#if !XFS_BIG_INUMS
208
	mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
209
#endif
210

211 212 213 214 215
	/*
	 * These can be overridden by the mount option parsing.
	 */
	mp->m_logbufs = -1;
	mp->m_logbsize = -1;
216 217 218 219 220 221 222 223 224 225 226 227

	if (!options)
		goto done;

	while ((this_char = strsep(&options, ",")) != NULL) {
		if (!*this_char)
			continue;
		if ((value = strchr(this_char, '=')) != NULL)
			*value++ = 0;

		if (!strcmp(this_char, MNTOPT_LOGBUFS)) {
			if (!value || !*value) {
228
				xfs_warn(mp, "%s option requires an argument",
229 230 231
					this_char);
				return EINVAL;
			}
232 233
			if (kstrtoint(value, 10, &mp->m_logbufs))
				return EINVAL;
234 235
		} else if (!strcmp(this_char, MNTOPT_LOGBSIZE)) {
			if (!value || !*value) {
236
				xfs_warn(mp, "%s option requires an argument",
237 238 239
					this_char);
				return EINVAL;
			}
240 241
			if (suffix_kstrtoint(value, 10, &mp->m_logbsize))
				return EINVAL;
242 243
		} else if (!strcmp(this_char, MNTOPT_LOGDEV)) {
			if (!value || !*value) {
244
				xfs_warn(mp, "%s option requires an argument",
245 246 247
					this_char);
				return EINVAL;
			}
248 249 250
			mp->m_logname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
			if (!mp->m_logname)
				return ENOMEM;
251
		} else if (!strcmp(this_char, MNTOPT_MTPT)) {
252
			xfs_warn(mp, "%s option not allowed on this system",
C
Christoph Hellwig 已提交
253 254
				this_char);
			return EINVAL;
255 256
		} else if (!strcmp(this_char, MNTOPT_RTDEV)) {
			if (!value || !*value) {
257
				xfs_warn(mp, "%s option requires an argument",
258 259 260
					this_char);
				return EINVAL;
			}
261 262 263
			mp->m_rtname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
			if (!mp->m_rtname)
				return ENOMEM;
264 265
		} else if (!strcmp(this_char, MNTOPT_BIOSIZE)) {
			if (!value || !*value) {
266
				xfs_warn(mp, "%s option requires an argument",
267 268 269
					this_char);
				return EINVAL;
			}
270 271
			if (kstrtoint(value, 10, &iosize))
				return EINVAL;
C
Christoph Hellwig 已提交
272
			iosizelog = ffs(iosize) - 1;
273 274
		} else if (!strcmp(this_char, MNTOPT_ALLOCSIZE)) {
			if (!value || !*value) {
275
				xfs_warn(mp, "%s option requires an argument",
276 277 278
					this_char);
				return EINVAL;
			}
279 280
			if (suffix_kstrtoint(value, 10, &iosize))
				return EINVAL;
281
			iosizelog = ffs(iosize) - 1;
282 283 284 285 286 287 288
		} else if (!strcmp(this_char, MNTOPT_GRPID) ||
			   !strcmp(this_char, MNTOPT_BSDGROUPS)) {
			mp->m_flags |= XFS_MOUNT_GRPID;
		} else if (!strcmp(this_char, MNTOPT_NOGRPID) ||
			   !strcmp(this_char, MNTOPT_SYSVGROUPS)) {
			mp->m_flags &= ~XFS_MOUNT_GRPID;
		} else if (!strcmp(this_char, MNTOPT_WSYNC)) {
289
			mp->m_flags |= XFS_MOUNT_WSYNC;
290
		} else if (!strcmp(this_char, MNTOPT_NORECOVERY)) {
291
			mp->m_flags |= XFS_MOUNT_NORECOVERY;
292
		} else if (!strcmp(this_char, MNTOPT_NOALIGN)) {
293
			mp->m_flags |= XFS_MOUNT_NOALIGN;
294
		} else if (!strcmp(this_char, MNTOPT_SWALLOC)) {
295
			mp->m_flags |= XFS_MOUNT_SWALLOC;
296 297
		} else if (!strcmp(this_char, MNTOPT_SUNIT)) {
			if (!value || !*value) {
298
				xfs_warn(mp, "%s option requires an argument",
299 300 301
					this_char);
				return EINVAL;
			}
302 303
			if (kstrtoint(value, 10, &dsunit))
				return EINVAL;
304 305
		} else if (!strcmp(this_char, MNTOPT_SWIDTH)) {
			if (!value || !*value) {
306
				xfs_warn(mp, "%s option requires an argument",
307 308 309
					this_char);
				return EINVAL;
			}
310 311
			if (kstrtoint(value, 10, &dswidth))
				return EINVAL;
312 313
		} else if (!strcmp(this_char, MNTOPT_32BITINODE)) {
			mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
314
		} else if (!strcmp(this_char, MNTOPT_64BITINODE)) {
315
			mp->m_flags &= ~XFS_MOUNT_SMALL_INUMS;
316
#if !XFS_BIG_INUMS
317
			xfs_warn(mp, "%s option not allowed on this system",
318 319 320 321
				this_char);
			return EINVAL;
#endif
		} else if (!strcmp(this_char, MNTOPT_NOUUID)) {
322
			mp->m_flags |= XFS_MOUNT_NOUUID;
323
		} else if (!strcmp(this_char, MNTOPT_BARRIER)) {
324
			mp->m_flags |= XFS_MOUNT_BARRIER;
325
		} else if (!strcmp(this_char, MNTOPT_NOBARRIER)) {
326
			mp->m_flags &= ~XFS_MOUNT_BARRIER;
327
		} else if (!strcmp(this_char, MNTOPT_IKEEP)) {
328
			mp->m_flags |= XFS_MOUNT_IKEEP;
329
		} else if (!strcmp(this_char, MNTOPT_NOIKEEP)) {
330
			mp->m_flags &= ~XFS_MOUNT_IKEEP;
331
		} else if (!strcmp(this_char, MNTOPT_LARGEIO)) {
332
			mp->m_flags &= ~XFS_MOUNT_COMPAT_IOSIZE;
333
		} else if (!strcmp(this_char, MNTOPT_NOLARGEIO)) {
334
			mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
335
		} else if (!strcmp(this_char, MNTOPT_ATTR2)) {
336
			mp->m_flags |= XFS_MOUNT_ATTR2;
337
		} else if (!strcmp(this_char, MNTOPT_NOATTR2)) {
338 339
			mp->m_flags &= ~XFS_MOUNT_ATTR2;
			mp->m_flags |= XFS_MOUNT_NOATTR2;
340
		} else if (!strcmp(this_char, MNTOPT_FILESTREAM)) {
341
			mp->m_flags |= XFS_MOUNT_FILESTREAMS;
342
		} else if (!strcmp(this_char, MNTOPT_NOQUOTA)) {
343 344 345
			mp->m_qflags &= ~XFS_ALL_QUOTA_ACCT;
			mp->m_qflags &= ~XFS_ALL_QUOTA_ENFD;
			mp->m_qflags &= ~XFS_ALL_QUOTA_ACTIVE;
346 347 348
		} else if (!strcmp(this_char, MNTOPT_QUOTA) ||
			   !strcmp(this_char, MNTOPT_UQUOTA) ||
			   !strcmp(this_char, MNTOPT_USRQUOTA)) {
349 350
			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
					 XFS_UQUOTA_ENFD);
351 352
		} else if (!strcmp(this_char, MNTOPT_QUOTANOENF) ||
			   !strcmp(this_char, MNTOPT_UQUOTANOENF)) {
353 354
			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE);
			mp->m_qflags &= ~XFS_UQUOTA_ENFD;
355 356
		} else if (!strcmp(this_char, MNTOPT_PQUOTA) ||
			   !strcmp(this_char, MNTOPT_PRJQUOTA)) {
357
			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
358
					 XFS_PQUOTA_ENFD);
359
		} else if (!strcmp(this_char, MNTOPT_PQUOTANOENF)) {
360
			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE);
361
			mp->m_qflags &= ~XFS_PQUOTA_ENFD;
362 363
		} else if (!strcmp(this_char, MNTOPT_GQUOTA) ||
			   !strcmp(this_char, MNTOPT_GRPQUOTA)) {
364
			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
365
					 XFS_GQUOTA_ENFD);
366
		} else if (!strcmp(this_char, MNTOPT_GQUOTANOENF)) {
367
			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE);
368
			mp->m_qflags &= ~XFS_GQUOTA_ENFD;
369
		} else if (!strcmp(this_char, MNTOPT_DELAYLOG)) {
370 371
			xfs_warn(mp,
	"delaylog is the default now, option is deprecated.");
372
		} else if (!strcmp(this_char, MNTOPT_NODELAYLOG)) {
373
			xfs_warn(mp,
374
	"nodelaylog support has been removed, option is deprecated.");
375 376 377 378
		} else if (!strcmp(this_char, MNTOPT_DISCARD)) {
			mp->m_flags |= XFS_MOUNT_DISCARD;
		} else if (!strcmp(this_char, MNTOPT_NODISCARD)) {
			mp->m_flags &= ~XFS_MOUNT_DISCARD;
379
		} else if (!strcmp(this_char, "ihashsize")) {
380 381
			xfs_warn(mp,
	"ihashsize no longer used, option is deprecated.");
382
		} else if (!strcmp(this_char, "osyncisdsync")) {
383 384
			xfs_warn(mp,
	"osyncisdsync has no effect, option is deprecated.");
385
		} else if (!strcmp(this_char, "osyncisosync")) {
386 387
			xfs_warn(mp,
	"osyncisosync has no effect, option is deprecated.");
388
		} else if (!strcmp(this_char, "irixsgid")) {
389 390
			xfs_warn(mp,
	"irixsgid is now a sysctl(2) variable, option is deprecated.");
391
		} else {
392
			xfs_warn(mp, "unknown mount option [%s].", this_char);
393 394 395 396
			return EINVAL;
		}
	}

397 398 399 400 401
	/*
	 * no recovery flag requires a read-only mount
	 */
	if ((mp->m_flags & XFS_MOUNT_NORECOVERY) &&
	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
402
		xfs_warn(mp, "no-recovery mounts must be read-only.");
403
		return EINVAL;
404 405
	}

406
	if ((mp->m_flags & XFS_MOUNT_NOALIGN) && (dsunit || dswidth)) {
407 408
		xfs_warn(mp,
	"sunit and swidth options incompatible with the noalign option");
409 410 411
		return EINVAL;
	}

C
Christoph Hellwig 已提交
412 413
#ifndef CONFIG_XFS_QUOTA
	if (XFS_IS_QUOTA_RUNNING(mp)) {
414
		xfs_warn(mp, "quota support not available in this kernel.");
C
Christoph Hellwig 已提交
415 416 417 418
		return EINVAL;
	}
#endif

419
	if ((dsunit && !dswidth) || (!dsunit && dswidth)) {
420
		xfs_warn(mp, "sunit and swidth must be specified together");
421 422 423 424
		return EINVAL;
	}

	if (dsunit && (dswidth % dsunit != 0)) {
425 426
		xfs_warn(mp,
	"stripe width (%d) must be a multiple of the stripe unit (%d)",
427 428 429 430
			dswidth, dsunit);
		return EINVAL;
	}

431
done:
J
Jie Liu 已提交
432
	if (dsunit && !(mp->m_flags & XFS_MOUNT_NOALIGN)) {
433 434 435 436 437 438
		/*
		 * At this point the superblock has not been read
		 * in, therefore we do not know the block size.
		 * Before the mount call ends we will convert
		 * these to FSBs.
		 */
J
Jie Liu 已提交
439 440
		mp->m_dalign = dsunit;
		mp->m_swidth = dswidth;
441 442 443 444 445 446
	}

	if (mp->m_logbufs != -1 &&
	    mp->m_logbufs != 0 &&
	    (mp->m_logbufs < XLOG_MIN_ICLOGS ||
	     mp->m_logbufs > XLOG_MAX_ICLOGS)) {
447
		xfs_warn(mp, "invalid logbufs value: %d [not %d-%d]",
448 449 450 451 452 453 454 455
			mp->m_logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
		return XFS_ERROR(EINVAL);
	}
	if (mp->m_logbsize != -1 &&
	    mp->m_logbsize !=  0 &&
	    (mp->m_logbsize < XLOG_MIN_RECORD_BSIZE ||
	     mp->m_logbsize > XLOG_MAX_RECORD_BSIZE ||
	     !is_power_of_2(mp->m_logbsize))) {
456 457
		xfs_warn(mp,
			"invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
458 459 460 461 462 463 464
			mp->m_logbsize);
		return XFS_ERROR(EINVAL);
	}

	if (iosizelog) {
		if (iosizelog > XFS_MAX_IO_LOG ||
		    iosizelog < XFS_MIN_IO_LOG) {
465
			xfs_warn(mp, "invalid log iosize: %d [not %d-%d]",
466 467 468 469 470 471 472 473
				iosizelog, XFS_MIN_IO_LOG,
				XFS_MAX_IO_LOG);
			return XFS_ERROR(EINVAL);
		}

		mp->m_flags |= XFS_MOUNT_DFLT_IOSIZE;
		mp->m_readio_log = iosizelog;
		mp->m_writeio_log = iosizelog;
474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490
	}

	return 0;
}

struct proc_xfs_info {
	int	flag;
	char	*str;
};

STATIC int
xfs_showargs(
	struct xfs_mount	*mp,
	struct seq_file		*m)
{
	static struct proc_xfs_info xfs_info_set[] = {
		/* the few simple ones we can get from the mount struct */
491
		{ XFS_MOUNT_IKEEP,		"," MNTOPT_IKEEP },
492 493 494 495 496 497 498 499
		{ XFS_MOUNT_WSYNC,		"," MNTOPT_WSYNC },
		{ XFS_MOUNT_NOALIGN,		"," MNTOPT_NOALIGN },
		{ XFS_MOUNT_SWALLOC,		"," MNTOPT_SWALLOC },
		{ XFS_MOUNT_NOUUID,		"," MNTOPT_NOUUID },
		{ XFS_MOUNT_NORECOVERY,		"," MNTOPT_NORECOVERY },
		{ XFS_MOUNT_ATTR2,		"," MNTOPT_ATTR2 },
		{ XFS_MOUNT_FILESTREAMS,	"," MNTOPT_FILESTREAM },
		{ XFS_MOUNT_GRPID,		"," MNTOPT_GRPID },
500
		{ XFS_MOUNT_DISCARD,		"," MNTOPT_DISCARD },
501
		{ XFS_MOUNT_SMALL_INUMS,	"," MNTOPT_32BITINODE },
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
		{ 0, NULL }
	};
	static struct proc_xfs_info xfs_info_unset[] = {
		/* the few simple ones we can get from the mount struct */
		{ XFS_MOUNT_COMPAT_IOSIZE,	"," MNTOPT_LARGEIO },
		{ XFS_MOUNT_BARRIER,		"," MNTOPT_NOBARRIER },
		{ XFS_MOUNT_SMALL_INUMS,	"," MNTOPT_64BITINODE },
		{ 0, NULL }
	};
	struct proc_xfs_info	*xfs_infop;

	for (xfs_infop = xfs_info_set; xfs_infop->flag; xfs_infop++) {
		if (mp->m_flags & xfs_infop->flag)
			seq_puts(m, xfs_infop->str);
	}
	for (xfs_infop = xfs_info_unset; xfs_infop->flag; xfs_infop++) {
		if (!(mp->m_flags & xfs_infop->flag))
			seq_puts(m, xfs_infop->str);
	}

	if (mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)
		seq_printf(m, "," MNTOPT_ALLOCSIZE "=%dk",
				(int)(1 << mp->m_writeio_log) >> 10);

	if (mp->m_logbufs > 0)
		seq_printf(m, "," MNTOPT_LOGBUFS "=%d", mp->m_logbufs);
	if (mp->m_logbsize > 0)
		seq_printf(m, "," MNTOPT_LOGBSIZE "=%dk", mp->m_logbsize >> 10);

	if (mp->m_logname)
		seq_printf(m, "," MNTOPT_LOGDEV "=%s", mp->m_logname);
	if (mp->m_rtname)
		seq_printf(m, "," MNTOPT_RTDEV "=%s", mp->m_rtname);

	if (mp->m_dalign > 0)
		seq_printf(m, "," MNTOPT_SUNIT "=%d",
				(int)XFS_FSB_TO_BB(mp, mp->m_dalign));
	if (mp->m_swidth > 0)
		seq_printf(m, "," MNTOPT_SWIDTH "=%d",
				(int)XFS_FSB_TO_BB(mp, mp->m_swidth));

	if (mp->m_qflags & (XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD))
		seq_puts(m, "," MNTOPT_USRQUOTA);
	else if (mp->m_qflags & XFS_UQUOTA_ACCT)
		seq_puts(m, "," MNTOPT_UQUOTANOENF);

548
	if (mp->m_qflags & XFS_PQUOTA_ACCT) {
549
		if (mp->m_qflags & XFS_PQUOTA_ENFD)
550 551 552
			seq_puts(m, "," MNTOPT_PRJQUOTA);
		else
			seq_puts(m, "," MNTOPT_PQUOTANOENF);
553 554
	}
	if (mp->m_qflags & XFS_GQUOTA_ACCT) {
555
		if (mp->m_qflags & XFS_GQUOTA_ENFD)
556 557 558 559
			seq_puts(m, "," MNTOPT_GRPQUOTA);
		else
			seq_puts(m, "," MNTOPT_GQUOTANOENF);
	}
560 561 562 563 564 565

	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
		seq_puts(m, "," MNTOPT_NOQUOTA);

	return 0;
}
L
Linus Torvalds 已提交
566 567 568 569 570 571 572 573 574
__uint64_t
xfs_max_file_offset(
	unsigned int		blockshift)
{
	unsigned int		pagefactor = 1;
	unsigned int		bitshift = BITS_PER_LONG - 1;

	/* Figure out maximum filesize, on Linux this can depend on
	 * the filesystem blocksize (on 32 bit platforms).
C
Christoph Hellwig 已提交
575
	 * __block_write_begin does this in an [unsigned] long...
L
Linus Torvalds 已提交
576 577 578 579 580 581 582 583 584 585 586 587
	 *      page->index << (PAGE_CACHE_SHIFT - bbits)
	 * So, for page sized blocks (4K on 32 bit platforms),
	 * this wraps at around 8Tb (hence MAX_LFS_FILESIZE which is
	 *      (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
	 * but for smaller blocksizes it is less (bbits = log2 bsize).
	 * Note1: get_block_t takes a long (implicit cast from above)
	 * Note2: The Large Block Device (LBD and HAVE_SECTOR_T) patch
	 * can optionally convert the [unsigned] long from above into
	 * an [unsigned] long long.
	 */

#if BITS_PER_LONG == 32
588
# if defined(CONFIG_LBDAF)
L
Linus Torvalds 已提交
589 590 591 592 593 594 595 596 597 598 599
	ASSERT(sizeof(sector_t) == 8);
	pagefactor = PAGE_CACHE_SIZE;
	bitshift = BITS_PER_LONG;
# else
	pagefactor = PAGE_CACHE_SIZE >> (PAGE_CACHE_SHIFT - blockshift);
# endif
#endif

	return (((__uint64_t)pagefactor) << bitshift) - 1;
}

600 601 602 603
xfs_agnumber_t
xfs_set_inode32(struct xfs_mount *mp)
{
	xfs_agnumber_t	index = 0;
604
	xfs_agnumber_t	maxagi = 0;
605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
	xfs_sb_t	*sbp = &mp->m_sb;
	xfs_agnumber_t	max_metadata;
	xfs_agino_t	agino =	XFS_OFFBNO_TO_AGINO(mp, sbp->sb_agblocks -1, 0);
	xfs_ino_t	ino = XFS_AGINO_TO_INO(mp, sbp->sb_agcount -1, agino);
	xfs_perag_t	*pag;

	/* Calculate how much should be reserved for inodes to meet
	 * the max inode percentage.
	 */
	if (mp->m_maxicount) {
		__uint64_t	icount;

		icount = sbp->sb_dblocks * sbp->sb_imax_pct;
		do_div(icount, 100);
		icount += sbp->sb_agblocks - 1;
		do_div(icount, sbp->sb_agblocks);
		max_metadata = icount;
	} else {
		max_metadata = sbp->sb_agcount;
	}

	for (index = 0; index < sbp->sb_agcount; index++) {
		ino = XFS_AGINO_TO_INO(mp, index, agino);
628

629
		if (ino > XFS_MAXINUMBER_32) {
630 631 632 633 634
			pag = xfs_perag_get(mp, index);
			pag->pagi_inodeok = 0;
			pag->pagf_metadata = 0;
			xfs_perag_put(pag);
			continue;
635 636 637 638
		}

		pag = xfs_perag_get(mp, index);
		pag->pagi_inodeok = 1;
639
		maxagi++;
640 641 642 643
		if (index < max_metadata)
			pag->pagf_metadata = 1;
		xfs_perag_put(pag);
	}
644 645 646 647
	mp->m_flags |= (XFS_MOUNT_32BITINODES |
			XFS_MOUNT_SMALL_INUMS);

	return maxagi;
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673
}

xfs_agnumber_t
xfs_set_inode64(struct xfs_mount *mp)
{
	xfs_agnumber_t index = 0;

	for (index = 0; index < mp->m_sb.sb_agcount; index++) {
		struct xfs_perag	*pag;

		pag = xfs_perag_get(mp, index);
		pag->pagi_inodeok = 1;
		pag->pagf_metadata = 0;
		xfs_perag_put(pag);
	}

	/* There is no need for lock protection on m_flags,
	 * the rw_semaphore of the VFS superblock is locked
	 * during mount/umount/remount operations, so this is
	 * enough to avoid concurency on the m_flags field
	 */
	mp->m_flags &= ~(XFS_MOUNT_32BITINODES |
			 XFS_MOUNT_SMALL_INUMS);
	return index;
}

H
Hannes Eder 已提交
674
STATIC int
L
Linus Torvalds 已提交
675 676 677 678 679 680 681
xfs_blkdev_get(
	xfs_mount_t		*mp,
	const char		*name,
	struct block_device	**bdevp)
{
	int			error = 0;

682 683
	*bdevp = blkdev_get_by_path(name, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
				    mp);
L
Linus Torvalds 已提交
684 685
	if (IS_ERR(*bdevp)) {
		error = PTR_ERR(*bdevp);
686
		xfs_warn(mp, "Invalid device [%s], error=%d\n", name, error);
L
Linus Torvalds 已提交
687 688 689 690 691
	}

	return -error;
}

H
Hannes Eder 已提交
692
STATIC void
L
Linus Torvalds 已提交
693 694 695 696
xfs_blkdev_put(
	struct block_device	*bdev)
{
	if (bdev)
697
		blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
L
Linus Torvalds 已提交
698 699
}

700 701 702 703
void
xfs_blkdev_issue_flush(
	xfs_buftarg_t		*buftarg)
{
704
	blkdev_issue_flush(buftarg->bt_bdev, GFP_NOFS, NULL);
705
}
L
Linus Torvalds 已提交
706

707 708 709 710 711
STATIC void
xfs_close_devices(
	struct xfs_mount	*mp)
{
	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
712
		struct block_device *logdev = mp->m_logdev_targp->bt_bdev;
713
		xfs_free_buftarg(mp, mp->m_logdev_targp);
714
		xfs_blkdev_put(logdev);
715 716
	}
	if (mp->m_rtdev_targp) {
717
		struct block_device *rtdev = mp->m_rtdev_targp->bt_bdev;
718
		xfs_free_buftarg(mp, mp->m_rtdev_targp);
719
		xfs_blkdev_put(rtdev);
720
	}
721
	xfs_free_buftarg(mp, mp->m_ddev_targp);
722 723 724 725 726 727 728 729 730 731 732 733 734 735
}

/*
 * The file system configurations are:
 *	(1) device (partition) with data and internal log
 *	(2) logical volume with data and log subvolumes.
 *	(3) logical volume with data, log, and realtime subvolumes.
 *
 * We only have to handle opening the log and realtime volumes here if
 * they are present.  The data subvolume has already been opened by
 * get_sb_bdev() and is stored in sb->s_bdev.
 */
STATIC int
xfs_open_devices(
736
	struct xfs_mount	*mp)
737 738 739 740 741 742 743 744
{
	struct block_device	*ddev = mp->m_super->s_bdev;
	struct block_device	*logdev = NULL, *rtdev = NULL;
	int			error;

	/*
	 * Open real time and log devices - order is important.
	 */
745 746
	if (mp->m_logname) {
		error = xfs_blkdev_get(mp, mp->m_logname, &logdev);
747 748 749 750
		if (error)
			goto out;
	}

751 752
	if (mp->m_rtname) {
		error = xfs_blkdev_get(mp, mp->m_rtname, &rtdev);
753 754 755 756
		if (error)
			goto out_close_logdev;

		if (rtdev == ddev || rtdev == logdev) {
757 758
			xfs_warn(mp,
	"Cannot mount filesystem with identical rtdev and ddev/logdev.");
759 760 761 762 763 764 765 766 767
			error = EINVAL;
			goto out_close_rtdev;
		}
	}

	/*
	 * Setup xfs_mount buffer target pointers
	 */
	error = ENOMEM;
768
	mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev, 0, mp->m_fsname);
769 770 771 772
	if (!mp->m_ddev_targp)
		goto out_close_rtdev;

	if (rtdev) {
773 774
		mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev, 1,
							mp->m_fsname);
775 776 777 778 779
		if (!mp->m_rtdev_targp)
			goto out_free_ddev_targ;
	}

	if (logdev && logdev != ddev) {
780 781
		mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev, 1,
							mp->m_fsname);
782 783 784 785 786 787 788 789 790 791
		if (!mp->m_logdev_targp)
			goto out_free_rtdev_targ;
	} else {
		mp->m_logdev_targp = mp->m_ddev_targp;
	}

	return 0;

 out_free_rtdev_targ:
	if (mp->m_rtdev_targp)
792
		xfs_free_buftarg(mp, mp->m_rtdev_targp);
793
 out_free_ddev_targ:
794
	xfs_free_buftarg(mp, mp->m_ddev_targp);
795 796 797 798 799 800 801 802 803 804
 out_close_rtdev:
	if (rtdev)
		xfs_blkdev_put(rtdev);
 out_close_logdev:
	if (logdev && logdev != ddev)
		xfs_blkdev_put(logdev);
 out:
	return error;
}

805 806 807 808 809 810 811 812
/*
 * Setup xfs_mount buffer target pointers based on superblock
 */
STATIC int
xfs_setup_devices(
	struct xfs_mount	*mp)
{
	int			error;
813

814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839
	error = xfs_setsize_buftarg(mp->m_ddev_targp, mp->m_sb.sb_blocksize,
				    mp->m_sb.sb_sectsize);
	if (error)
		return error;

	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
		unsigned int	log_sector_size = BBSIZE;

		if (xfs_sb_version_hassector(&mp->m_sb))
			log_sector_size = mp->m_sb.sb_logsectsize;
		error = xfs_setsize_buftarg(mp->m_logdev_targp,
					    mp->m_sb.sb_blocksize,
					    log_sector_size);
		if (error)
			return error;
	}
	if (mp->m_rtdev_targp) {
		error = xfs_setsize_buftarg(mp->m_rtdev_targp,
					    mp->m_sb.sb_blocksize,
					    mp->m_sb.sb_sectsize);
		if (error)
			return error;
	}

	return 0;
}
840

841 842 843 844 845 846 847 848 849 850 851 852 853 854
STATIC int
xfs_init_mount_workqueues(
	struct xfs_mount	*mp)
{
	mp->m_data_workqueue = alloc_workqueue("xfs-data/%s",
			WQ_MEM_RECLAIM, 0, mp->m_fsname);
	if (!mp->m_data_workqueue)
		goto out;

	mp->m_unwritten_workqueue = alloc_workqueue("xfs-conv/%s",
			WQ_MEM_RECLAIM, 0, mp->m_fsname);
	if (!mp->m_unwritten_workqueue)
		goto out_destroy_data_iodone_queue;

855 856 857 858
	mp->m_cil_workqueue = alloc_workqueue("xfs-cil/%s",
			WQ_MEM_RECLAIM, 0, mp->m_fsname);
	if (!mp->m_cil_workqueue)
		goto out_destroy_unwritten;
D
Dave Chinner 已提交
859 860

	mp->m_reclaim_workqueue = alloc_workqueue("xfs-reclaim/%s",
861
			0, 0, mp->m_fsname);
D
Dave Chinner 已提交
862 863 864 865
	if (!mp->m_reclaim_workqueue)
		goto out_destroy_cil;

	mp->m_log_workqueue = alloc_workqueue("xfs-log/%s",
866
			0, 0, mp->m_fsname);
D
Dave Chinner 已提交
867 868 869
	if (!mp->m_log_workqueue)
		goto out_destroy_reclaim;

870
	mp->m_eofblocks_workqueue = alloc_workqueue("xfs-eofblocks/%s",
871
			0, 0, mp->m_fsname);
872 873 874
	if (!mp->m_eofblocks_workqueue)
		goto out_destroy_log;

875 876
	return 0;

877 878
out_destroy_log:
	destroy_workqueue(mp->m_log_workqueue);
D
Dave Chinner 已提交
879 880 881 882
out_destroy_reclaim:
	destroy_workqueue(mp->m_reclaim_workqueue);
out_destroy_cil:
	destroy_workqueue(mp->m_cil_workqueue);
883 884
out_destroy_unwritten:
	destroy_workqueue(mp->m_unwritten_workqueue);
885 886 887 888 889 890 891 892 893 894
out_destroy_data_iodone_queue:
	destroy_workqueue(mp->m_data_workqueue);
out:
	return -ENOMEM;
}

STATIC void
xfs_destroy_mount_workqueues(
	struct xfs_mount	*mp)
{
895
	destroy_workqueue(mp->m_eofblocks_workqueue);
D
Dave Chinner 已提交
896 897
	destroy_workqueue(mp->m_log_workqueue);
	destroy_workqueue(mp->m_reclaim_workqueue);
898
	destroy_workqueue(mp->m_cil_workqueue);
899 900 901 902
	destroy_workqueue(mp->m_data_workqueue);
	destroy_workqueue(mp->m_unwritten_workqueue);
}

D
Dave Chinner 已提交
903 904 905 906 907 908 909 910 911 912 913 914 915
/*
 * Flush all dirty data to disk. Must not be called while holding an XFS_ILOCK
 * or a page lock. We use sync_inodes_sb() here to ensure we block while waiting
 * for IO to complete so that we effectively throttle multiple callers to the
 * rate at which IO is completing.
 */
void
xfs_flush_inodes(
	struct xfs_mount	*mp)
{
	struct super_block	*sb = mp->m_super;

	if (down_read_trylock(&sb->s_umount)) {
916
		sync_inodes_sb(sb);
D
Dave Chinner 已提交
917 918 919 920
		up_read(&sb->s_umount);
	}
}

921
/* Catch misguided souls that try to use this interface on XFS */
L
Linus Torvalds 已提交
922
STATIC struct inode *
923
xfs_fs_alloc_inode(
L
Linus Torvalds 已提交
924 925
	struct super_block	*sb)
{
926
	BUG();
927
	return NULL;
L
Linus Torvalds 已提交
928 929
}

930
/*
931 932
 * Now that the generic code is guaranteed not to be accessing
 * the linux inode, we can reclaim the inode.
933
 */
L
Linus Torvalds 已提交
934
STATIC void
935
xfs_fs_destroy_inode(
C
Christoph Hellwig 已提交
936
	struct inode		*inode)
L
Linus Torvalds 已提交
937
{
C
Christoph Hellwig 已提交
938 939
	struct xfs_inode	*ip = XFS_I(inode);

C
Christoph Hellwig 已提交
940
	trace_xfs_destroy_inode(ip);
941 942

	XFS_STATS_INC(vn_reclaim);
C
Christoph Hellwig 已提交
943 944 945 946 947 948 949 950 951 952

	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);

	/*
	 * We should never get here with one of the reclaim flags already set.
	 */
	ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIMABLE));
	ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIM));

	/*
953 954 955 956 957
	 * We always use background reclaim here because even if the
	 * inode is clean, it still may be under IO and hence we have
	 * to take the flush lock. The background reclaim path handles
	 * this more efficiently than we can here, so simply let background
	 * reclaim tear down all inodes.
C
Christoph Hellwig 已提交
958
	 */
959
	xfs_inode_set_reclaim_tag(ip);
L
Linus Torvalds 已提交
960 961
}

962 963 964 965
/*
 * Slab object creation initialisation for the XFS inode.
 * This covers only the idempotent fields in the XFS inode;
 * all other fields need to be initialised on allocation
966
 * from the slab. This avoids the need to repeatedly initialise
967 968 969
 * fields in the xfs inode that left in the initialise state
 * when freeing the inode.
 */
970 971
STATIC void
xfs_fs_inode_init_once(
972 973 974 975 976
	void			*inode)
{
	struct xfs_inode	*ip = inode;

	memset(ip, 0, sizeof(struct xfs_inode));
977 978 979 980 981

	/* vfs inode */
	inode_init_once(VFS_I(ip));

	/* xfs inode */
982 983 984 985 986 987 988
	atomic_set(&ip->i_pincount, 0);
	spin_lock_init(&ip->i_flags_lock);

	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
		     "xfsino", ip->i_ino);
}

L
Linus Torvalds 已提交
989
STATIC void
990
xfs_fs_evict_inode(
L
Linus Torvalds 已提交
991 992
	struct inode		*inode)
{
993
	xfs_inode_t		*ip = XFS_I(inode);
994

995 996
	ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));

997
	trace_xfs_evict_inode(ip);
C
Christoph Hellwig 已提交
998

999
	truncate_inode_pages(&inode->i_data, 0);
1000
	clear_inode(inode);
1001 1002 1003 1004 1005
	XFS_STATS_INC(vn_rele);
	XFS_STATS_INC(vn_remove);
	XFS_STATS_DEC(vn_active);

	xfs_inactive(ip);
1006
}
L
Linus Torvalds 已提交
1007

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
/*
 * We do an unlocked check for XFS_IDONTCACHE here because we are already
 * serialised against cache hits here via the inode->i_lock and igrab() in
 * xfs_iget_cache_hit(). Hence a lookup that might clear this flag will not be
 * racing with us, and it avoids needing to grab a spinlock here for every inode
 * we drop the final reference on.
 */
STATIC int
xfs_fs_drop_inode(
	struct inode		*inode)
{
	struct xfs_inode	*ip = XFS_I(inode);

	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
}

1024 1025 1026 1027 1028 1029 1030 1031 1032
STATIC void
xfs_free_fsname(
	struct xfs_mount	*mp)
{
	kfree(mp->m_fsname);
	kfree(mp->m_rtname);
	kfree(mp->m_logname);
}

L
Linus Torvalds 已提交
1033
STATIC void
1034
xfs_fs_put_super(
L
Linus Torvalds 已提交
1035 1036
	struct super_block	*sb)
{
1037
	struct xfs_mount	*mp = XFS_M(sb);
L
Linus Torvalds 已提交
1038

D
Dave Chinner 已提交
1039 1040 1041
	xfs_filestream_unmount(mp);
	xfs_unmountfs(mp);

1042
	xfs_freesb(mp);
C
Christoph Hellwig 已提交
1043
	xfs_icsb_destroy_counters(mp);
1044
	xfs_destroy_mount_workqueues(mp);
1045
	xfs_close_devices(mp);
1046
	xfs_free_fsname(mp);
C
Christoph Hellwig 已提交
1047
	kfree(mp);
L
Linus Torvalds 已提交
1048 1049 1050
}

STATIC int
C
Christoph Hellwig 已提交
1051
xfs_fs_sync_fs(
L
Linus Torvalds 已提交
1052 1053 1054
	struct super_block	*sb,
	int			wait)
{
1055
	struct xfs_mount	*mp = XFS_M(sb);
L
Linus Torvalds 已提交
1056

1057
	/*
C
Christoph Hellwig 已提交
1058
	 * Doing anything during the async pass would be counterproductive.
1059
	 */
C
Christoph Hellwig 已提交
1060
	if (!wait)
C
Christoph Hellwig 已提交
1061 1062
		return 0;

D
Dave Chinner 已提交
1063
	xfs_log_force(mp, XFS_LOG_SYNC);
C
Christoph Hellwig 已提交
1064
	if (laptop_mode) {
L
Linus Torvalds 已提交
1065 1066
		/*
		 * The disk must be active because we're syncing.
1067
		 * We schedule log work now (now that the disk is
L
Linus Torvalds 已提交
1068 1069
		 * active) instead of later (when it might not be).
		 */
1070
		flush_delayed_work(&mp->m_log->l_work);
L
Linus Torvalds 已提交
1071 1072
	}

C
Christoph Hellwig 已提交
1073
	return 0;
L
Linus Torvalds 已提交
1074 1075 1076
}

STATIC int
1077
xfs_fs_statfs(
1078
	struct dentry		*dentry,
L
Linus Torvalds 已提交
1079 1080
	struct kstatfs		*statp)
{
C
Christoph Hellwig 已提交
1081 1082
	struct xfs_mount	*mp = XFS_M(dentry->d_sb);
	xfs_sb_t		*sbp = &mp->m_sb;
C
Christoph Hellwig 已提交
1083
	struct xfs_inode	*ip = XFS_I(dentry->d_inode);
C
Christoph Hellwig 已提交
1084 1085
	__uint64_t		fakeinos, id;
	xfs_extlen_t		lsize;
1086
	__int64_t		ffree;
C
Christoph Hellwig 已提交
1087 1088 1089 1090 1091 1092 1093 1094

	statp->f_type = XFS_SB_MAGIC;
	statp->f_namelen = MAXNAMELEN - 1;

	id = huge_encode_dev(mp->m_ddev_targp->bt_dev);
	statp->f_fsid.val[0] = (u32)id;
	statp->f_fsid.val[1] = (u32)(id >> 32);

1095
	xfs_icsb_sync_counters(mp, XFS_ICSB_LAZY_COUNT);
C
Christoph Hellwig 已提交
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106

	spin_lock(&mp->m_sb_lock);
	statp->f_bsize = sbp->sb_blocksize;
	lsize = sbp->sb_logstart ? sbp->sb_logblocks : 0;
	statp->f_blocks = sbp->sb_dblocks - lsize;
	statp->f_bfree = statp->f_bavail =
				sbp->sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
	fakeinos = statp->f_bfree << sbp->sb_inopblog;
	statp->f_files =
	    MIN(sbp->sb_icount + fakeinos, (__uint64_t)XFS_MAXINUMBER);
	if (mp->m_maxicount)
C
Christoph Hellwig 已提交
1107 1108 1109
		statp->f_files = min_t(typeof(statp->f_files),
					statp->f_files,
					mp->m_maxicount);
1110 1111 1112 1113 1114

	/* make sure statp->f_ffree does not underflow */
	ffree = statp->f_files - (sbp->sb_icount - sbp->sb_ifree);
	statp->f_ffree = max_t(__int64_t, ffree, 0);

C
Christoph Hellwig 已提交
1115 1116
	spin_unlock(&mp->m_sb_lock);

1117
	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
1118 1119
	    ((mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))) ==
			      (XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))
C
Christoph Hellwig 已提交
1120
		xfs_qm_statvfs(ip, statp);
C
Christoph Hellwig 已提交
1121
	return 0;
L
Linus Torvalds 已提交
1122 1123
}

E
Eric Sandeen 已提交
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
STATIC void
xfs_save_resvblks(struct xfs_mount *mp)
{
	__uint64_t resblks = 0;

	mp->m_resblks_save = mp->m_resblks;
	xfs_reserve_blocks(mp, &resblks, NULL);
}

STATIC void
xfs_restore_resvblks(struct xfs_mount *mp)
{
	__uint64_t resblks;

	if (mp->m_resblks_save) {
		resblks = mp->m_resblks_save;
		mp->m_resblks_save = 0;
	} else
		resblks = xfs_default_resblks(mp);

	xfs_reserve_blocks(mp, &resblks, NULL);
}

1147 1148 1149 1150 1151
/*
 * Trigger writeback of all the dirty metadata in the file system.
 *
 * This ensures that the metadata is written to their location on disk rather
 * than just existing in transactions in the log. This means after a quiesce
1152 1153
 * there is no log replay required to write the inodes to disk - this is the
 * primary difference between a sync and a quiesce.
1154
 *
1155 1156
 * Note: xfs_log_quiesce() stops background log work - the callers must ensure
 * it is started again when appropriate.
1157
 */
1158
static void
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
xfs_quiesce_attr(
	struct xfs_mount	*mp)
{
	int	error = 0;

	/* wait for all modifications to complete */
	while (atomic_read(&mp->m_active_trans) > 0)
		delay(100);

	/* force the log to unpin objects from the now complete transactions */
	xfs_log_force(mp, XFS_LOG_SYNC);

	/* reclaim inodes to do any IO before the freeze completes */
	xfs_reclaim_inodes(mp, 0);
	xfs_reclaim_inodes(mp, SYNC_WAIT);

	/* Push the superblock and write an unmount record */
	error = xfs_log_sbcount(mp);
	if (error)
		xfs_warn(mp, "xfs_attr_quiesce: failed to log sb changes. "
				"Frozen image may not be consistent.");
	/*
1181 1182
	 * Just warn here till VFS can correctly support
	 * read-only remount without racing.
1183
	 */
1184
	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
1185

1186
	xfs_log_quiesce(mp);
1187 1188
}

L
Linus Torvalds 已提交
1189
STATIC int
1190
xfs_fs_remount(
L
Linus Torvalds 已提交
1191 1192 1193 1194
	struct super_block	*sb,
	int			*flags,
	char			*options)
{
1195
	struct xfs_mount	*mp = XFS_M(sb);
1196 1197
	substring_t		args[MAX_OPT_ARGS];
	char			*p;
1198
	int			error;
L
Linus Torvalds 已提交
1199

1200 1201
	while ((p = strsep(&options, ",")) != NULL) {
		int token;
1202

1203 1204
		if (!*p)
			continue;
1205

1206 1207 1208
		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_barrier:
1209
			mp->m_flags |= XFS_MOUNT_BARRIER;
1210 1211
			break;
		case Opt_nobarrier:
1212
			mp->m_flags &= ~XFS_MOUNT_BARRIER;
1213
			break;
1214
		case Opt_inode64:
1215
			mp->m_maxagi = xfs_set_inode64(mp);
1216
			break;
1217 1218 1219
		case Opt_inode32:
			mp->m_maxagi = xfs_set_inode32(mp);
			break;
1220
		default:
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
			/*
			 * Logically we would return an error here to prevent
			 * users from believing they might have changed
			 * mount options using remount which can't be changed.
			 *
			 * But unfortunately mount(8) adds all options from
			 * mtab and fstab to the mount arguments in some cases
			 * so we can't blindly reject options, but have to
			 * check for each specified option if it actually
			 * differs from the currently set option and only
			 * reject it if that's the case.
			 *
			 * Until that is implemented we return success for
			 * every remount request, and silently ignore all
			 * options that we can't actually change.
			 */
#if 0
1238
			xfs_info(mp,
1239
		"mount option \"%s\" not supported for remount", p);
1240
			return -EINVAL;
1241
#else
1242
			break;
1243
#endif
1244
		}
1245 1246
	}

1247
	/* ro -> rw */
1248 1249
	if ((mp->m_flags & XFS_MOUNT_RDONLY) && !(*flags & MS_RDONLY)) {
		mp->m_flags &= ~XFS_MOUNT_RDONLY;
1250 1251 1252 1253 1254 1255 1256 1257

		/*
		 * If this is the first remount to writeable state we
		 * might have some superblock changes to update.
		 */
		if (mp->m_update_flags) {
			error = xfs_mount_log_sb(mp, mp->m_update_flags);
			if (error) {
1258
				xfs_warn(mp, "failed to write sb changes");
1259 1260 1261 1262
				return error;
			}
			mp->m_update_flags = 0;
		}
1263 1264 1265 1266 1267

		/*
		 * Fill out the reserve pool if it is empty. Use the stashed
		 * value if it is non-zero, otherwise go with the default.
		 */
E
Eric Sandeen 已提交
1268
		xfs_restore_resvblks(mp);
1269
		xfs_log_work_queue(mp);
1270 1271 1272 1273
	}

	/* rw -> ro */
	if (!(mp->m_flags & XFS_MOUNT_RDONLY) && (*flags & MS_RDONLY)) {
1274
		/*
D
Dave Chinner 已提交
1275 1276 1277 1278 1279
		 * Before we sync the metadata, we need to free up the reserve
		 * block pool so that the used block count in the superblock on
		 * disk is correct at the end of the remount. Stash the current
		 * reserve pool size so that if we get remounted rw, we can
		 * return it to the same size.
1280
		 */
E
Eric Sandeen 已提交
1281
		xfs_save_resvblks(mp);
D
David Chinner 已提交
1282
		xfs_quiesce_attr(mp);
1283 1284 1285
		mp->m_flags |= XFS_MOUNT_RDONLY;
	}

1286
	return 0;
L
Linus Torvalds 已提交
1287 1288
}

C
Christoph Hellwig 已提交
1289 1290
/*
 * Second stage of a freeze. The data is already frozen so we only
D
David Chinner 已提交
1291
 * need to take care of the metadata. Once that's done write a dummy
C
Christoph Hellwig 已提交
1292 1293
 * record to dirty the log in case of a crash while frozen.
 */
1294 1295
STATIC int
xfs_fs_freeze(
L
Linus Torvalds 已提交
1296 1297
	struct super_block	*sb)
{
C
Christoph Hellwig 已提交
1298 1299
	struct xfs_mount	*mp = XFS_M(sb);

E
Eric Sandeen 已提交
1300
	xfs_save_resvblks(mp);
D
David Chinner 已提交
1301
	xfs_quiesce_attr(mp);
1302
	return -xfs_fs_log_dummy(mp);
L
Linus Torvalds 已提交
1303 1304
}

E
Eric Sandeen 已提交
1305 1306 1307 1308 1309 1310 1311
STATIC int
xfs_fs_unfreeze(
	struct super_block	*sb)
{
	struct xfs_mount	*mp = XFS_M(sb);

	xfs_restore_resvblks(mp);
1312
	xfs_log_work_queue(mp);
E
Eric Sandeen 已提交
1313 1314 1315
	return 0;
}

L
Linus Torvalds 已提交
1316
STATIC int
1317
xfs_fs_show_options(
L
Linus Torvalds 已提交
1318
	struct seq_file		*m,
1319
	struct dentry		*root)
L
Linus Torvalds 已提交
1320
{
1321
	return -xfs_showargs(XFS_M(root->d_sb), m);
L
Linus Torvalds 已提交
1322 1323
}

1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
/*
 * This function fills in xfs_mount_t fields based on mount args.
 * Note: the superblock _has_ now been read in.
 */
STATIC int
xfs_finish_flags(
	struct xfs_mount	*mp)
{
	int			ronly = (mp->m_flags & XFS_MOUNT_RDONLY);

1334
	/* Fail a mount where the logbuf is smaller than the log stripe */
1335
	if (xfs_sb_version_haslogv2(&mp->m_sb)) {
1336 1337
		if (mp->m_logbsize <= 0 &&
		    mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE) {
1338
			mp->m_logbsize = mp->m_sb.sb_logsunit;
1339 1340
		} else if (mp->m_logbsize > 0 &&
			   mp->m_logbsize < mp->m_sb.sb_logsunit) {
1341 1342
			xfs_warn(mp,
		"logbuf size must be greater than or equal to log stripe size");
1343 1344 1345 1346
			return XFS_ERROR(EINVAL);
		}
	} else {
		/* Fail a mount if the logbuf is larger than 32K */
1347
		if (mp->m_logbsize > XLOG_BIG_RECORD_BSIZE) {
1348 1349
			xfs_warn(mp,
		"logbuf size for version 1 logs must be 16K or 32K");
1350 1351 1352 1353
			return XFS_ERROR(EINVAL);
		}
	}

1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
	/*
	 * V5 filesystems always use attr2 format for attributes.
	 */
	if (xfs_sb_version_hascrc(&mp->m_sb) &&
	    (mp->m_flags & XFS_MOUNT_NOATTR2)) {
		xfs_warn(mp,
"Cannot mount a V5 filesystem as %s. %s is always enabled for V5 filesystems.",
			MNTOPT_NOATTR2, MNTOPT_ATTR2);
		return XFS_ERROR(EINVAL);
	}

1365 1366 1367 1368 1369
	/*
	 * mkfs'ed attr2 will turn on attr2 mount unless explicitly
	 * told by noattr2 to turn it off
	 */
	if (xfs_sb_version_hasattr2(&mp->m_sb) &&
1370
	    !(mp->m_flags & XFS_MOUNT_NOATTR2))
1371 1372 1373 1374 1375 1376
		mp->m_flags |= XFS_MOUNT_ATTR2;

	/*
	 * prohibit r/w mounts of read-only filesystems
	 */
	if ((mp->m_sb.sb_flags & XFS_SBF_READONLY) && !ronly) {
1377 1378
		xfs_warn(mp,
			"cannot mount a read-only filesystem as read-write");
1379 1380 1381
		return XFS_ERROR(EROFS);
	}

1382 1383 1384 1385 1386 1387 1388 1389
	if ((mp->m_qflags & (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE)) &&
	    (mp->m_qflags & (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE)) &&
	    !xfs_sb_version_has_pquotino(&mp->m_sb)) {
		xfs_warn(mp,
		  "Super block does not support project and group quota together");
		return XFS_ERROR(EINVAL);
	}

1390 1391 1392
	return 0;
}

L
Linus Torvalds 已提交
1393
STATIC int
1394
xfs_fs_fill_super(
L
Linus Torvalds 已提交
1395 1396 1397 1398
	struct super_block	*sb,
	void			*data,
	int			silent)
{
1399
	struct inode		*root;
1400
	struct xfs_mount	*mp = NULL;
C
Christoph Hellwig 已提交
1401
	int			flags = 0, error = ENOMEM;
1402

C
Christoph Hellwig 已提交
1403 1404
	mp = kzalloc(sizeof(struct xfs_mount), GFP_KERNEL);
	if (!mp)
1405
		goto out;
L
Linus Torvalds 已提交
1406

C
Christoph Hellwig 已提交
1407 1408 1409
	spin_lock_init(&mp->m_sb_lock);
	mutex_init(&mp->m_growlock);
	atomic_set(&mp->m_active_trans, 0);
D
Dave Chinner 已提交
1410
	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
1411
	INIT_DELAYED_WORK(&mp->m_eofblocks_work, xfs_eofblocks_worker);
1412

C
Christoph Hellwig 已提交
1413 1414
	mp->m_super = sb;
	sb->s_fs_info = mp;
L
Linus Torvalds 已提交
1415

C
Christoph Hellwig 已提交
1416
	error = xfs_parseargs(mp, (char *)data);
1417
	if (error)
1418
		goto out_free_fsname;
L
Linus Torvalds 已提交
1419 1420

	sb_min_blocksize(sb, BBSIZE);
1421
	sb->s_xattr = xfs_xattr_handlers;
1422
	sb->s_export_op = &xfs_export_operations;
1423
#ifdef CONFIG_XFS_QUOTA
1424
	sb->s_qcop = &xfs_quotactl_operations;
1425
#endif
1426
	sb->s_op = &xfs_super_operations;
L
Linus Torvalds 已提交
1427

1428
	if (silent)
1429 1430
		flags |= XFS_MFSI_QUIET;

1431
	error = xfs_open_devices(mp);
1432
	if (error)
C
Christoph Hellwig 已提交
1433
		goto out_free_fsname;
1434

1435
	error = xfs_init_mount_workqueues(mp);
1436 1437
	if (error)
		goto out_close_devices;
C
Christoph Hellwig 已提交
1438

1439 1440 1441 1442
	error = xfs_icsb_init_counters(mp);
	if (error)
		goto out_destroy_workqueues;

1443 1444
	error = xfs_readsb(mp, flags);
	if (error)
1445 1446 1447
		goto out_destroy_counters;

	error = xfs_finish_flags(mp);
1448
	if (error)
1449
		goto out_free_sb;
1450

1451
	error = xfs_setup_devices(mp);
1452
	if (error)
1453
		goto out_free_sb;
1454 1455 1456

	error = xfs_filestream_mount(mp);
	if (error)
1457
		goto out_free_sb;
1458

1459 1460 1461 1462
	/*
	 * we must configure the block size in the superblock before we run the
	 * full mount process as the mount process can lookup and cache inodes.
	 */
C
Christoph Hellwig 已提交
1463 1464 1465
	sb->s_magic = XFS_SB_MAGIC;
	sb->s_blocksize = mp->m_sb.sb_blocksize;
	sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1;
L
Linus Torvalds 已提交
1466
	sb->s_maxbytes = xfs_max_file_offset(sb->s_blocksize_bits);
1467
	sb->s_max_links = XFS_MAXLINK;
L
Linus Torvalds 已提交
1468 1469 1470
	sb->s_time_gran = 1;
	set_posix_acl_flag(sb);

D
Dave Chinner 已提交
1471 1472 1473 1474
	/* version 5 superblocks support inode version counters. */
	if (XFS_SB_VERSION_NUM(&mp->m_sb) == XFS_SB_VERSION_5)
		sb->s_flags |= MS_I_VERSION;

1475
	error = xfs_mountfs(mp);
1476
	if (error)
D
Dave Chinner 已提交
1477
		goto out_filestream_unmount;
1478

1479
	root = igrab(VFS_I(mp->m_rootip));
1480
	if (!root) {
C
Christoph Hellwig 已提交
1481
		error = ENOENT;
1482
		goto out_unmount;
C
Christoph Hellwig 已提交
1483
	}
1484
	sb->s_root = d_make_root(root);
1485 1486
	if (!sb->s_root) {
		error = ENOMEM;
1487
		goto out_unmount;
L
Linus Torvalds 已提交
1488
	}
1489

L
Linus Torvalds 已提交
1490
	return 0;
D
Dave Chinner 已提交
1491

D
Dave Chinner 已提交
1492
 out_filestream_unmount:
1493
	xfs_filestream_unmount(mp);
1494 1495
 out_free_sb:
	xfs_freesb(mp);
1496
 out_destroy_counters:
C
Christoph Hellwig 已提交
1497
	xfs_icsb_destroy_counters(mp);
1498 1499
out_destroy_workqueues:
	xfs_destroy_mount_workqueues(mp);
1500
 out_close_devices:
1501
	xfs_close_devices(mp);
1502 1503
 out_free_fsname:
	xfs_free_fsname(mp);
C
Christoph Hellwig 已提交
1504
	kfree(mp);
1505
 out:
C
Christoph Hellwig 已提交
1506
	return -error;
1507

1508
 out_unmount:
1509
	xfs_filestream_unmount(mp);
1510
	xfs_unmountfs(mp);
1511
	goto out_free_sb;
L
Linus Torvalds 已提交
1512 1513
}

A
Al Viro 已提交
1514 1515
STATIC struct dentry *
xfs_fs_mount(
L
Linus Torvalds 已提交
1516 1517 1518
	struct file_system_type	*fs_type,
	int			flags,
	const char		*dev_name,
A
Al Viro 已提交
1519
	void			*data)
L
Linus Torvalds 已提交
1520
{
A
Al Viro 已提交
1521
	return mount_bdev(fs_type, flags, dev_name, data, xfs_fs_fill_super);
1522 1523
}

1524
static long
1525
xfs_fs_nr_cached_objects(
1526 1527
	struct super_block	*sb,
	int			nid)
1528 1529 1530 1531
{
	return xfs_reclaim_inodes_count(XFS_M(sb));
}

1532
static long
1533 1534
xfs_fs_free_cached_objects(
	struct super_block	*sb,
1535 1536
	long			nr_to_scan,
	int			nid)
1537
{
1538
	return xfs_reclaim_inodes_nr(XFS_M(sb), nr_to_scan);
1539 1540
}

1541
static const struct super_operations xfs_super_operations = {
1542 1543
	.alloc_inode		= xfs_fs_alloc_inode,
	.destroy_inode		= xfs_fs_destroy_inode,
1544
	.evict_inode		= xfs_fs_evict_inode,
1545
	.drop_inode		= xfs_fs_drop_inode,
1546
	.put_super		= xfs_fs_put_super,
C
Christoph Hellwig 已提交
1547
	.sync_fs		= xfs_fs_sync_fs,
1548
	.freeze_fs		= xfs_fs_freeze,
E
Eric Sandeen 已提交
1549
	.unfreeze_fs		= xfs_fs_unfreeze,
1550 1551 1552
	.statfs			= xfs_fs_statfs,
	.remount_fs		= xfs_fs_remount,
	.show_options		= xfs_fs_show_options,
1553 1554
	.nr_cached_objects	= xfs_fs_nr_cached_objects,
	.free_cached_objects	= xfs_fs_free_cached_objects,
L
Linus Torvalds 已提交
1555 1556
};

A
Andrew Morton 已提交
1557
static struct file_system_type xfs_fs_type = {
L
Linus Torvalds 已提交
1558 1559
	.owner			= THIS_MODULE,
	.name			= "xfs",
A
Al Viro 已提交
1560
	.mount			= xfs_fs_mount,
L
Linus Torvalds 已提交
1561 1562 1563
	.kill_sb		= kill_block_super,
	.fs_flags		= FS_REQUIRES_DEV,
};
1564
MODULE_ALIAS_FS("xfs");
L
Linus Torvalds 已提交
1565

1566 1567 1568 1569 1570 1571
STATIC int __init
xfs_init_zones(void)
{

	xfs_ioend_zone = kmem_zone_init(sizeof(xfs_ioend_t), "xfs_ioend");
	if (!xfs_ioend_zone)
1572
		goto out;
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587

	xfs_ioend_pool = mempool_create_slab_pool(4 * MAX_BUF_PER_PAGE,
						  xfs_ioend_zone);
	if (!xfs_ioend_pool)
		goto out_destroy_ioend_zone;

	xfs_log_ticket_zone = kmem_zone_init(sizeof(xlog_ticket_t),
						"xfs_log_ticket");
	if (!xfs_log_ticket_zone)
		goto out_destroy_ioend_pool;

	xfs_bmap_free_item_zone = kmem_zone_init(sizeof(xfs_bmap_free_item_t),
						"xfs_bmap_free_item");
	if (!xfs_bmap_free_item_zone)
		goto out_destroy_log_ticket_zone;
1588

1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	xfs_btree_cur_zone = kmem_zone_init(sizeof(xfs_btree_cur_t),
						"xfs_btree_cur");
	if (!xfs_btree_cur_zone)
		goto out_destroy_bmap_free_item_zone;

	xfs_da_state_zone = kmem_zone_init(sizeof(xfs_da_state_t),
						"xfs_da_state");
	if (!xfs_da_state_zone)
		goto out_destroy_btree_cur_zone;

	xfs_ifork_zone = kmem_zone_init(sizeof(xfs_ifork_t), "xfs_ifork");
	if (!xfs_ifork_zone)
1601
		goto out_destroy_da_state_zone;
1602 1603 1604 1605 1606

	xfs_trans_zone = kmem_zone_init(sizeof(xfs_trans_t), "xfs_trans");
	if (!xfs_trans_zone)
		goto out_destroy_ifork_zone;

1607 1608 1609 1610 1611 1612
	xfs_log_item_desc_zone =
		kmem_zone_init(sizeof(struct xfs_log_item_desc),
			       "xfs_log_item_desc");
	if (!xfs_log_item_desc_zone)
		goto out_destroy_trans_zone;

1613 1614 1615 1616 1617
	/*
	 * The size of the zone allocated buf log item is the maximum
	 * size possible under XFS.  This wastes a little bit of memory,
	 * but it is much faster.
	 */
1618 1619
	xfs_buf_item_zone = kmem_zone_init(sizeof(struct xfs_buf_log_item),
					   "xfs_buf_item");
1620
	if (!xfs_buf_item_zone)
1621
		goto out_destroy_log_item_desc_zone;
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636

	xfs_efd_zone = kmem_zone_init((sizeof(xfs_efd_log_item_t) +
			((XFS_EFD_MAX_FAST_EXTENTS - 1) *
				 sizeof(xfs_extent_t))), "xfs_efd_item");
	if (!xfs_efd_zone)
		goto out_destroy_buf_item_zone;

	xfs_efi_zone = kmem_zone_init((sizeof(xfs_efi_log_item_t) +
			((XFS_EFI_MAX_FAST_EXTENTS - 1) *
				sizeof(xfs_extent_t))), "xfs_efi_item");
	if (!xfs_efi_zone)
		goto out_destroy_efd_zone;

	xfs_inode_zone =
		kmem_zone_init_flags(sizeof(xfs_inode_t), "xfs_inode",
1637 1638
			KM_ZONE_HWALIGN | KM_ZONE_RECLAIM | KM_ZONE_SPREAD,
			xfs_fs_inode_init_once);
1639 1640 1641 1642 1643 1644 1645 1646
	if (!xfs_inode_zone)
		goto out_destroy_efi_zone;

	xfs_ili_zone =
		kmem_zone_init_flags(sizeof(xfs_inode_log_item_t), "xfs_ili",
					KM_ZONE_SPREAD, NULL);
	if (!xfs_ili_zone)
		goto out_destroy_inode_zone;
D
Dave Chinner 已提交
1647 1648 1649 1650
	xfs_icreate_zone = kmem_zone_init(sizeof(struct xfs_icreate_item),
					"xfs_icr");
	if (!xfs_icreate_zone)
		goto out_destroy_ili_zone;
1651 1652 1653

	return 0;

D
Dave Chinner 已提交
1654 1655
 out_destroy_ili_zone:
	kmem_zone_destroy(xfs_ili_zone);
1656 1657 1658 1659 1660 1661 1662 1663
 out_destroy_inode_zone:
	kmem_zone_destroy(xfs_inode_zone);
 out_destroy_efi_zone:
	kmem_zone_destroy(xfs_efi_zone);
 out_destroy_efd_zone:
	kmem_zone_destroy(xfs_efd_zone);
 out_destroy_buf_item_zone:
	kmem_zone_destroy(xfs_buf_item_zone);
1664 1665
 out_destroy_log_item_desc_zone:
	kmem_zone_destroy(xfs_log_item_desc_zone);
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
 out_destroy_trans_zone:
	kmem_zone_destroy(xfs_trans_zone);
 out_destroy_ifork_zone:
	kmem_zone_destroy(xfs_ifork_zone);
 out_destroy_da_state_zone:
	kmem_zone_destroy(xfs_da_state_zone);
 out_destroy_btree_cur_zone:
	kmem_zone_destroy(xfs_btree_cur_zone);
 out_destroy_bmap_free_item_zone:
	kmem_zone_destroy(xfs_bmap_free_item_zone);
 out_destroy_log_ticket_zone:
	kmem_zone_destroy(xfs_log_ticket_zone);
 out_destroy_ioend_pool:
	mempool_destroy(xfs_ioend_pool);
 out_destroy_ioend_zone:
	kmem_zone_destroy(xfs_ioend_zone);
 out:
	return -ENOMEM;
}

STATIC void
xfs_destroy_zones(void)
{
1689 1690 1691 1692 1693
	/*
	 * Make sure all delayed rcu free are flushed before we
	 * destroy caches.
	 */
	rcu_barrier();
D
Dave Chinner 已提交
1694
	kmem_zone_destroy(xfs_icreate_zone);
1695 1696 1697 1698 1699
	kmem_zone_destroy(xfs_ili_zone);
	kmem_zone_destroy(xfs_inode_zone);
	kmem_zone_destroy(xfs_efi_zone);
	kmem_zone_destroy(xfs_efd_zone);
	kmem_zone_destroy(xfs_buf_item_zone);
1700
	kmem_zone_destroy(xfs_log_item_desc_zone);
1701 1702 1703 1704 1705 1706 1707 1708 1709 1710
	kmem_zone_destroy(xfs_trans_zone);
	kmem_zone_destroy(xfs_ifork_zone);
	kmem_zone_destroy(xfs_da_state_zone);
	kmem_zone_destroy(xfs_btree_cur_zone);
	kmem_zone_destroy(xfs_bmap_free_item_zone);
	kmem_zone_destroy(xfs_log_ticket_zone);
	mempool_destroy(xfs_ioend_pool);
	kmem_zone_destroy(xfs_ioend_zone);

}
L
Linus Torvalds 已提交
1711

1712 1713 1714
STATIC int __init
xfs_init_workqueues(void)
{
1715 1716 1717 1718 1719 1720 1721 1722
	/*
	 * The allocation workqueue can be used in memory reclaim situations
	 * (writepage path), and parallelism is only limited by the number of
	 * AGs in all the filesystems mounted. Hence use the default large
	 * max_active value for this workqueue.
	 */
	xfs_alloc_wq = alloc_workqueue("xfsalloc", WQ_MEM_RECLAIM, 0);
	if (!xfs_alloc_wq)
D
Dave Chinner 已提交
1723
		return -ENOMEM;
1724

1725 1726 1727
	return 0;
}

1728
STATIC void
1729 1730
xfs_destroy_workqueues(void)
{
1731
	destroy_workqueue(xfs_alloc_wq);
1732 1733
}

L
Linus Torvalds 已提交
1734
STATIC int __init
1735
init_xfs_fs(void)
L
Linus Torvalds 已提交
1736 1737 1738
{
	int			error;

1739 1740
	printk(KERN_INFO XFS_VERSION_STRING " with "
			 XFS_BUILD_OPTIONS " enabled\n");
L
Linus Torvalds 已提交
1741

1742
	xfs_dir_startup();
L
Linus Torvalds 已提交
1743

1744
	error = xfs_init_zones();
1745 1746 1747
	if (error)
		goto out;

1748
	error = xfs_init_workqueues();
1749
	if (error)
C
Christoph Hellwig 已提交
1750
		goto out_destroy_zones;
1751

1752 1753 1754 1755
	error = xfs_mru_cache_init();
	if (error)
		goto out_destroy_wq;

1756 1757 1758
	error = xfs_filestream_init();
	if (error)
		goto out_mru_cache_uninit;
L
Linus Torvalds 已提交
1759

1760
	error = xfs_buf_init();
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770
	if (error)
		goto out_filestream_uninit;

	error = xfs_init_procfs();
	if (error)
		goto out_buf_terminate;

	error = xfs_sysctl_register();
	if (error)
		goto out_cleanup_procfs;
L
Linus Torvalds 已提交
1771

1772 1773 1774
	error = xfs_qm_init();
	if (error)
		goto out_sysctl_unregister;
L
Linus Torvalds 已提交
1775 1776 1777

	error = register_filesystem(&xfs_fs_type);
	if (error)
1778
		goto out_qm_exit;
L
Linus Torvalds 已提交
1779 1780
	return 0;

1781 1782
 out_qm_exit:
	xfs_qm_exit();
1783 1784 1785 1786 1787
 out_sysctl_unregister:
	xfs_sysctl_unregister();
 out_cleanup_procfs:
	xfs_cleanup_procfs();
 out_buf_terminate:
1788
	xfs_buf_terminate();
1789 1790 1791 1792
 out_filestream_uninit:
	xfs_filestream_uninit();
 out_mru_cache_uninit:
	xfs_mru_cache_uninit();
1793 1794
 out_destroy_wq:
	xfs_destroy_workqueues();
1795
 out_destroy_zones:
1796
	xfs_destroy_zones();
1797
 out:
L
Linus Torvalds 已提交
1798 1799 1800 1801
	return error;
}

STATIC void __exit
1802
exit_xfs_fs(void)
L
Linus Torvalds 已提交
1803
{
1804
	xfs_qm_exit();
L
Linus Torvalds 已提交
1805
	unregister_filesystem(&xfs_fs_type);
1806 1807
	xfs_sysctl_unregister();
	xfs_cleanup_procfs();
1808
	xfs_buf_terminate();
1809 1810
	xfs_filestream_uninit();
	xfs_mru_cache_uninit();
1811
	xfs_destroy_workqueues();
1812
	xfs_destroy_zones();
L
Linus Torvalds 已提交
1813 1814 1815 1816 1817 1818 1819 1820
}

module_init(init_xfs_fs);
module_exit(exit_xfs_fs);

MODULE_AUTHOR("Silicon Graphics, Inc.");
MODULE_DESCRIPTION(XFS_VERSION_STRING " with " XFS_BUILD_OPTIONS " enabled");
MODULE_LICENSE("GPL");