xfs_super.c 53.2 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"
L
Linus Torvalds 已提交
24 25
#include "xfs_sb.h"
#include "xfs_mount.h"
26
#include "xfs_da_format.h"
L
Linus Torvalds 已提交
27
#include "xfs_inode.h"
28
#include "xfs_btree.h"
L
Linus Torvalds 已提交
29
#include "xfs_bmap.h"
30
#include "xfs_alloc.h"
L
Linus Torvalds 已提交
31
#include "xfs_error.h"
C
Christoph Hellwig 已提交
32
#include "xfs_fsops.h"
33
#include "xfs_trans.h"
L
Linus Torvalds 已提交
34
#include "xfs_buf_item.h"
35
#include "xfs_log.h"
36
#include "xfs_log_priv.h"
37
#include "xfs_da_btree.h"
38
#include "xfs_dir2.h"
39 40 41
#include "xfs_extfree_item.h"
#include "xfs_mru_cache.h"
#include "xfs_inode_item.h"
42
#include "xfs_icache.h"
C
Christoph Hellwig 已提交
43
#include "xfs_trace.h"
D
Dave Chinner 已提交
44
#include "xfs_icreate_item.h"
45 46
#include "xfs_filestream.h"
#include "xfs_quota.h"
47
#include "xfs_sysfs.h"
48
#include "xfs_ondisk.h"
49
#include "xfs_rmap_item.h"
50
#include "xfs_refcount_item.h"
51
#include "xfs_bmap_item.h"
52
#include "xfs_reflink.h"
L
Linus Torvalds 已提交
53 54 55

#include <linux/namei.h>
#include <linux/init.h>
56
#include <linux/slab.h>
L
Linus Torvalds 已提交
57
#include <linux/mount.h>
58
#include <linux/mempool.h>
L
Linus Torvalds 已提交
59
#include <linux/writeback.h>
60
#include <linux/kthread.h>
61
#include <linux/freezer.h>
62
#include <linux/parser.h>
L
Linus Torvalds 已提交
63

64
static const struct super_operations xfs_super_operations;
65
struct bio_set *xfs_ioend_bioset;
66

D
Dave Chinner 已提交
67
static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
68 69 70
#ifdef DEBUG
static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
#endif
L
Linus Torvalds 已提交
71

72 73 74 75
/*
 * Table driven mount option parser.
 */
enum {
76 77 78 79 80 81 82 83 84 85
	Opt_logbufs, Opt_logbsize, Opt_logdev, Opt_rtdev, Opt_biosize,
	Opt_wsync, Opt_noalign, Opt_swalloc, Opt_sunit, Opt_swidth, Opt_nouuid,
	Opt_mtpt, Opt_grpid, Opt_nogrpid, Opt_bsdgroups, Opt_sysvgroups,
	Opt_allocsize, Opt_norecovery, Opt_barrier, Opt_nobarrier,
	Opt_inode64, Opt_inode32, Opt_ikeep, Opt_noikeep,
	Opt_largeio, Opt_nolargeio, Opt_attr2, Opt_noattr2, Opt_filestreams,
	Opt_quota, Opt_noquota, Opt_usrquota, Opt_grpquota, Opt_prjquota,
	Opt_uquota, Opt_gquota, Opt_pquota,
	Opt_uqnoenforce, Opt_gqnoenforce, Opt_pqnoenforce, Opt_qnoenforce,
	Opt_discard, Opt_nodiscard, Opt_dax, Opt_err,
86 87
};

88
static const match_table_t tokens = {
89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
	{Opt_logbufs,	"logbufs=%u"},	/* number of XFS log buffers */
	{Opt_logbsize,	"logbsize=%s"},	/* size of XFS log buffers */
	{Opt_logdev,	"logdev=%s"},	/* log device */
	{Opt_rtdev,	"rtdev=%s"},	/* realtime I/O device */
	{Opt_biosize,	"biosize=%u"},	/* log2 of preferred buffered io size */
	{Opt_wsync,	"wsync"},	/* safe-mode nfs compatible mount */
	{Opt_noalign,	"noalign"},	/* turn off stripe alignment */
	{Opt_swalloc,	"swalloc"},	/* turn on stripe width allocation */
	{Opt_sunit,	"sunit=%u"},	/* data volume stripe unit */
	{Opt_swidth,	"swidth=%u"},	/* data volume stripe width */
	{Opt_nouuid,	"nouuid"},	/* ignore filesystem UUID */
	{Opt_mtpt,	"mtpt"},	/* filesystem mount point */
	{Opt_grpid,	"grpid"},	/* group-ID from parent directory */
	{Opt_nogrpid,	"nogrpid"},	/* group-ID from current process */
	{Opt_bsdgroups,	"bsdgroups"},	/* group-ID from parent directory */
	{Opt_sysvgroups,"sysvgroups"},	/* group-ID from current process */
	{Opt_allocsize,	"allocsize=%s"},/* preferred allocation size */
	{Opt_norecovery,"norecovery"},	/* don't run XFS recovery */
	{Opt_inode64,	"inode64"},	/* inodes can be allocated anywhere */
	{Opt_inode32,   "inode32"},	/* inode allocation limited to
					 * XFS_MAXINUMBER_32 */
	{Opt_ikeep,	"ikeep"},	/* do not free empty inode clusters */
	{Opt_noikeep,	"noikeep"},	/* free empty inode clusters */
	{Opt_largeio,	"largeio"},	/* report large I/O sizes in stat() */
	{Opt_nolargeio,	"nolargeio"},	/* do not report large I/O sizes
					 * in stat(). */
	{Opt_attr2,	"attr2"},	/* do use attr2 attribute format */
	{Opt_noattr2,	"noattr2"},	/* do not use attr2 attribute format */
	{Opt_filestreams,"filestreams"},/* use filestreams allocator */
	{Opt_quota,	"quota"},	/* disk quotas (user) */
	{Opt_noquota,	"noquota"},	/* no quotas */
	{Opt_usrquota,	"usrquota"},	/* user quota enabled */
	{Opt_grpquota,	"grpquota"},	/* group quota enabled */
	{Opt_prjquota,	"prjquota"},	/* project quota enabled */
	{Opt_uquota,	"uquota"},	/* user quota (IRIX variant) */
	{Opt_gquota,	"gquota"},	/* group quota (IRIX variant) */
	{Opt_pquota,	"pquota"},	/* project quota (IRIX variant) */
	{Opt_uqnoenforce,"uqnoenforce"},/* user quota limit enforcement */
	{Opt_gqnoenforce,"gqnoenforce"},/* group quota limit enforcement */
	{Opt_pqnoenforce,"pqnoenforce"},/* project quota limit enforcement */
	{Opt_qnoenforce, "qnoenforce"},	/* same as uqnoenforce */
	{Opt_discard,	"discard"},	/* Discard unused blocks */
	{Opt_nodiscard,	"nodiscard"},	/* Do not discard unused blocks */

	{Opt_dax,	"dax"},		/* Enable direct access to bdev pages */
134 135 136 137 138 139

	/* Deprecated mount options scheduled for removal */
	{Opt_barrier,	"barrier"},	/* use writer barriers for log write and
					 * unwritten extent conversion */
	{Opt_nobarrier,	"nobarrier"},	/* .. disable */

140
	{Opt_err,	NULL},
141 142 143
};


144
STATIC int
145
suffix_kstrtoint(const substring_t *s, unsigned int base, int *res)
146
{
147
	int	last, shift_left_factor = 0, _res;
148 149 150 151 152 153
	char	*value;
	int	ret = 0;

	value = match_strdup(s);
	if (!value)
		return -ENOMEM;
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168

	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';
	}

169 170 171
	if (kstrtoint(value, base, &_res))
		ret = -EINVAL;
	kfree(value);
172
	*res = _res << shift_left_factor;
173
	return ret;
174 175
}

176 177 178 179 180 181
/*
 * 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.
E
Eric Sandeen 已提交
182 183 184 185
 *
 * *sb is const because this is also used to test options on the remount
 * path, and we don't want this to have any side effects at remount time.
 * Today this function does not change *sb, but just to future-proof...
186
 */
187 188 189
STATIC int
xfs_parseargs(
	struct xfs_mount	*mp,
C
Christoph Hellwig 已提交
190
	char			*options)
191
{
E
Eric Sandeen 已提交
192
	const struct super_block *sb = mp->m_super;
193 194
	char			*p;
	substring_t		args[MAX_OPT_ARGS];
195 196 197
	int			dsunit = 0;
	int			dswidth = 0;
	int			iosize = 0;
198
	__uint8_t		iosizelog = 0;
199

200 201 202 203 204 205
	/*
	 * 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)
D
Dave Chinner 已提交
206
		return -ENOMEM;
207 208
	mp->m_fsname_len = strlen(mp->m_fsname) + 1;

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224
	/*
	 * 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;
225

226 227 228 229 230
	/*
	 * These can be overridden by the mount option parsing.
	 */
	mp->m_logbufs = -1;
	mp->m_logbsize = -1;
231 232 233 234

	if (!options)
		goto done;

235 236 237 238
	while ((p = strsep(&options, ",")) != NULL) {
		int		token;

		if (!*p)
239 240
			continue;

241 242 243 244
		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_logbufs:
			if (match_int(args, &mp->m_logbufs))
D
Dave Chinner 已提交
245
				return -EINVAL;
246 247 248
			break;
		case Opt_logbsize:
			if (suffix_kstrtoint(args, 10, &mp->m_logbsize))
D
Dave Chinner 已提交
249
				return -EINVAL;
250 251 252
			break;
		case Opt_logdev:
			mp->m_logname = match_strdup(args);
253
			if (!mp->m_logname)
D
Dave Chinner 已提交
254
				return -ENOMEM;
255 256 257
			break;
		case Opt_mtpt:
			xfs_warn(mp, "%s option not allowed on this system", p);
D
Dave Chinner 已提交
258
			return -EINVAL;
259 260
		case Opt_rtdev:
			mp->m_rtname = match_strdup(args);
261
			if (!mp->m_rtname)
D
Dave Chinner 已提交
262
				return -ENOMEM;
263 264 265 266
			break;
		case Opt_allocsize:
		case Opt_biosize:
			if (suffix_kstrtoint(args, 10, &iosize))
D
Dave Chinner 已提交
267
				return -EINVAL;
268
			iosizelog = ffs(iosize) - 1;
269 270 271
			break;
		case Opt_grpid:
		case Opt_bsdgroups:
272
			mp->m_flags |= XFS_MOUNT_GRPID;
273 274 275
			break;
		case Opt_nogrpid:
		case Opt_sysvgroups:
276
			mp->m_flags &= ~XFS_MOUNT_GRPID;
277 278
			break;
		case Opt_wsync:
279
			mp->m_flags |= XFS_MOUNT_WSYNC;
280 281
			break;
		case Opt_norecovery:
282
			mp->m_flags |= XFS_MOUNT_NORECOVERY;
283 284
			break;
		case Opt_noalign:
285
			mp->m_flags |= XFS_MOUNT_NOALIGN;
286 287
			break;
		case Opt_swalloc:
288
			mp->m_flags |= XFS_MOUNT_SWALLOC;
289 290 291
			break;
		case Opt_sunit:
			if (match_int(args, &dsunit))
D
Dave Chinner 已提交
292
				return -EINVAL;
293 294 295
			break;
		case Opt_swidth:
			if (match_int(args, &dswidth))
D
Dave Chinner 已提交
296
				return -EINVAL;
297 298
			break;
		case Opt_inode32:
299
			mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
300 301
			break;
		case Opt_inode64:
302
			mp->m_flags &= ~XFS_MOUNT_SMALL_INUMS;
303 304
			break;
		case Opt_nouuid:
305
			mp->m_flags |= XFS_MOUNT_NOUUID;
306 307
			break;
		case Opt_ikeep:
308
			mp->m_flags |= XFS_MOUNT_IKEEP;
309 310
			break;
		case Opt_noikeep:
311
			mp->m_flags &= ~XFS_MOUNT_IKEEP;
312 313
			break;
		case Opt_largeio:
314
			mp->m_flags &= ~XFS_MOUNT_COMPAT_IOSIZE;
315 316
			break;
		case Opt_nolargeio:
317
			mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
318 319
			break;
		case Opt_attr2:
320
			mp->m_flags |= XFS_MOUNT_ATTR2;
321 322
			break;
		case Opt_noattr2:
323 324
			mp->m_flags &= ~XFS_MOUNT_ATTR2;
			mp->m_flags |= XFS_MOUNT_NOATTR2;
325 326
			break;
		case Opt_filestreams:
327
			mp->m_flags |= XFS_MOUNT_FILESTREAMS;
328 329
			break;
		case Opt_noquota:
330 331 332
			mp->m_qflags &= ~XFS_ALL_QUOTA_ACCT;
			mp->m_qflags &= ~XFS_ALL_QUOTA_ENFD;
			mp->m_qflags &= ~XFS_ALL_QUOTA_ACTIVE;
333 334 335 336
			break;
		case Opt_quota:
		case Opt_uquota:
		case Opt_usrquota:
337 338
			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
					 XFS_UQUOTA_ENFD);
339 340 341
			break;
		case Opt_qnoenforce:
		case Opt_uqnoenforce:
342 343
			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE);
			mp->m_qflags &= ~XFS_UQUOTA_ENFD;
344 345 346
			break;
		case Opt_pquota:
		case Opt_prjquota:
347
			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
348
					 XFS_PQUOTA_ENFD);
349 350
			break;
		case Opt_pqnoenforce:
351
			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE);
352
			mp->m_qflags &= ~XFS_PQUOTA_ENFD;
353
			break;
354 355
		case Opt_gquota:
		case Opt_grpquota:
356
			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
357
					 XFS_GQUOTA_ENFD);
358 359
			break;
		case Opt_gqnoenforce:
360
			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE);
361
			mp->m_qflags &= ~XFS_GQUOTA_ENFD;
362 363
			break;
		case Opt_discard:
364
			mp->m_flags |= XFS_MOUNT_DISCARD;
365 366
			break;
		case Opt_nodiscard:
367
			mp->m_flags &= ~XFS_MOUNT_DISCARD;
368
			break;
D
Dave Chinner 已提交
369
#ifdef CONFIG_FS_DAX
370
		case Opt_dax:
D
Dave Chinner 已提交
371
			mp->m_flags |= XFS_MOUNT_DAX;
372
			break;
D
Dave Chinner 已提交
373
#endif
374 375 376 377 378 379 380 381
		case Opt_barrier:
			xfs_warn(mp, "%s option is deprecated, ignoring.", p);
			mp->m_flags |= XFS_MOUNT_BARRIER;
			break;
		case Opt_nobarrier:
			xfs_warn(mp, "%s option is deprecated, ignoring.", p);
			mp->m_flags &= ~XFS_MOUNT_BARRIER;
			break;
382 383
		default:
			xfs_warn(mp, "unknown mount option [%s].", p);
D
Dave Chinner 已提交
384
			return -EINVAL;
385 386 387
		}
	}

388 389 390 391 392
	/*
	 * no recovery flag requires a read-only mount
	 */
	if ((mp->m_flags & XFS_MOUNT_NORECOVERY) &&
	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
393
		xfs_warn(mp, "no-recovery mounts must be read-only.");
D
Dave Chinner 已提交
394
		return -EINVAL;
395 396
	}

397
	if ((mp->m_flags & XFS_MOUNT_NOALIGN) && (dsunit || dswidth)) {
398 399
		xfs_warn(mp,
	"sunit and swidth options incompatible with the noalign option");
D
Dave Chinner 已提交
400
		return -EINVAL;
401 402
	}

C
Christoph Hellwig 已提交
403 404
#ifndef CONFIG_XFS_QUOTA
	if (XFS_IS_QUOTA_RUNNING(mp)) {
405
		xfs_warn(mp, "quota support not available in this kernel.");
D
Dave Chinner 已提交
406
		return -EINVAL;
C
Christoph Hellwig 已提交
407 408 409
	}
#endif

410
	if ((dsunit && !dswidth) || (!dsunit && dswidth)) {
411
		xfs_warn(mp, "sunit and swidth must be specified together");
D
Dave Chinner 已提交
412
		return -EINVAL;
413 414 415
	}

	if (dsunit && (dswidth % dsunit != 0)) {
416 417
		xfs_warn(mp,
	"stripe width (%d) must be a multiple of the stripe unit (%d)",
418
			dswidth, dsunit);
D
Dave Chinner 已提交
419
		return -EINVAL;
420 421
	}

422
done:
J
Jie Liu 已提交
423
	if (dsunit && !(mp->m_flags & XFS_MOUNT_NOALIGN)) {
424 425 426 427 428 429
		/*
		 * 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
430 431
		mp->m_dalign = dsunit;
		mp->m_swidth = dswidth;
432 433 434 435 436 437
	}

	if (mp->m_logbufs != -1 &&
	    mp->m_logbufs != 0 &&
	    (mp->m_logbufs < XLOG_MIN_ICLOGS ||
	     mp->m_logbufs > XLOG_MAX_ICLOGS)) {
438
		xfs_warn(mp, "invalid logbufs value: %d [not %d-%d]",
439
			mp->m_logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
D
Dave Chinner 已提交
440
		return -EINVAL;
441 442 443 444 445 446
	}
	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))) {
447 448
		xfs_warn(mp,
			"invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
449
			mp->m_logbsize);
D
Dave Chinner 已提交
450
		return -EINVAL;
451 452 453 454 455
	}

	if (iosizelog) {
		if (iosizelog > XFS_MAX_IO_LOG ||
		    iosizelog < XFS_MIN_IO_LOG) {
456
			xfs_warn(mp, "invalid log iosize: %d [not %d-%d]",
457 458
				iosizelog, XFS_MIN_IO_LOG,
				XFS_MAX_IO_LOG);
D
Dave Chinner 已提交
459
			return -EINVAL;
460 461 462 463 464
		}

		mp->m_flags |= XFS_MOUNT_DFLT_IOSIZE;
		mp->m_readio_log = iosizelog;
		mp->m_writeio_log = iosizelog;
465 466 467 468 469 470
	}

	return 0;
}

struct proc_xfs_info {
D
Dave Chinner 已提交
471 472
	uint64_t	flag;
	char		*str;
473 474 475 476 477 478 479 480 481
};

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 */
482 483 484 485 486 487 488 489 490 491 492 493
		{ XFS_MOUNT_IKEEP,		",ikeep" },
		{ XFS_MOUNT_WSYNC,		",wsync" },
		{ XFS_MOUNT_NOALIGN,		",noalign" },
		{ XFS_MOUNT_SWALLOC,		",swalloc" },
		{ XFS_MOUNT_NOUUID,		",nouuid" },
		{ XFS_MOUNT_NORECOVERY,		",norecovery" },
		{ XFS_MOUNT_ATTR2,		",attr2" },
		{ XFS_MOUNT_FILESTREAMS,	",filestreams" },
		{ XFS_MOUNT_GRPID,		",grpid" },
		{ XFS_MOUNT_DISCARD,		",discard" },
		{ XFS_MOUNT_SMALL_INUMS,	",inode32" },
		{ XFS_MOUNT_DAX,		",dax" },
494 495 496 497
		{ 0, NULL }
	};
	static struct proc_xfs_info xfs_info_unset[] = {
		/* the few simple ones we can get from the mount struct */
498 499 500
		{ XFS_MOUNT_COMPAT_IOSIZE,	",largeio" },
		{ XFS_MOUNT_BARRIER,		",nobarrier" },
		{ XFS_MOUNT_SMALL_INUMS,	",inode64" },
501 502 503 504 505 506 507 508 509 510 511 512 513 514
		{ 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)
515
		seq_printf(m, ",allocsize=%dk",
516 517 518
				(int)(1 << mp->m_writeio_log) >> 10);

	if (mp->m_logbufs > 0)
519
		seq_printf(m, ",logbufs=%d", mp->m_logbufs);
520
	if (mp->m_logbsize > 0)
521
		seq_printf(m, ",logbsize=%dk", mp->m_logbsize >> 10);
522 523

	if (mp->m_logname)
524
		seq_show_option(m, "logdev", mp->m_logname);
525
	if (mp->m_rtname)
526
		seq_show_option(m, "rtdev", mp->m_rtname);
527 528

	if (mp->m_dalign > 0)
529
		seq_printf(m, ",sunit=%d",
530 531
				(int)XFS_FSB_TO_BB(mp, mp->m_dalign));
	if (mp->m_swidth > 0)
532
		seq_printf(m, ",swidth=%d",
533 534 535
				(int)XFS_FSB_TO_BB(mp, mp->m_swidth));

	if (mp->m_qflags & (XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD))
536
		seq_puts(m, ",usrquota");
537
	else if (mp->m_qflags & XFS_UQUOTA_ACCT)
538
		seq_puts(m, ",uqnoenforce");
539

540
	if (mp->m_qflags & XFS_PQUOTA_ACCT) {
541
		if (mp->m_qflags & XFS_PQUOTA_ENFD)
542
			seq_puts(m, ",prjquota");
543
		else
544
			seq_puts(m, ",pqnoenforce");
545 546
	}
	if (mp->m_qflags & XFS_GQUOTA_ACCT) {
547
		if (mp->m_qflags & XFS_GQUOTA_ENFD)
548
			seq_puts(m, ",grpquota");
549
		else
550
			seq_puts(m, ",gqnoenforce");
551
	}
552 553

	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
554
		seq_puts(m, ",noquota");
555 556 557

	return 0;
}
558
static __uint64_t
L
Linus Torvalds 已提交
559 560 561 562 563 564 565 566
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
567
	 * __block_write_begin does this in an [unsigned] long...
568
	 *      page->index << (PAGE_SHIFT - bbits)
L
Linus Torvalds 已提交
569 570
	 * So, for page sized blocks (4K on 32 bit platforms),
	 * this wraps at around 8Tb (hence MAX_LFS_FILESIZE which is
571
	 *      (((u64)PAGE_SIZE << (BITS_PER_LONG-1))-1)
L
Linus Torvalds 已提交
572 573 574 575 576 577 578 579
	 * 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
580
# if defined(CONFIG_LBDAF)
L
Linus Torvalds 已提交
581
	ASSERT(sizeof(sector_t) == 8);
582
	pagefactor = PAGE_SIZE;
L
Linus Torvalds 已提交
583 584
	bitshift = BITS_PER_LONG;
# else
585
	pagefactor = PAGE_SIZE >> (PAGE_SHIFT - blockshift);
L
Linus Torvalds 已提交
586 587 588 589 590 591
# endif
#endif

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

592
/*
593 594 595 596 597 598 599 600 601 602 603 604 605
 * Set parameters for inode allocation heuristics, taking into account
 * filesystem size and inode32/inode64 mount options; i.e. specifically
 * whether or not XFS_MOUNT_SMALL_INUMS is set.
 *
 * Inode allocation patterns are altered only if inode32 is requested
 * (XFS_MOUNT_SMALL_INUMS), and the filesystem is sufficiently large.
 * If altered, XFS_MOUNT_32BITINODES is set as well.
 *
 * An agcount independent of that in the mount structure is provided
 * because in the growfs case, mp->m_sb.sb_agcount is not yet updated
 * to the potentially higher ag count.
 *
 * Returns the maximum AG index which may contain inodes.
606
 */
607
xfs_agnumber_t
608 609 610
xfs_set_inode_alloc(
	struct xfs_mount *mp,
	xfs_agnumber_t	agcount)
611
{
612
	xfs_agnumber_t	index;
613
	xfs_agnumber_t	maxagi = 0;
614 615
	xfs_sb_t	*sbp = &mp->m_sb;
	xfs_agnumber_t	max_metadata;
E
Eric Sandeen 已提交
616 617
	xfs_agino_t	agino;
	xfs_ino_t	ino;
618

619 620 621
	/*
	 * Calculate how much should be reserved for inodes to meet
	 * the max inode percentage.  Used only for inode32.
622 623 624 625 626 627 628 629 630 631
	 */
	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 {
632
		max_metadata = agcount;
633 634
	}

635
	/* Get the last possible inode in the filesystem */
E
Eric Sandeen 已提交
636
	agino =	XFS_OFFBNO_TO_AGINO(mp, sbp->sb_agblocks - 1, 0);
637 638 639 640 641 642 643 644 645 646 647
	ino = XFS_AGINO_TO_INO(mp, agcount - 1, agino);

	/*
	 * If user asked for no more than 32-bit inodes, and the fs is
	 * sufficiently large, set XFS_MOUNT_32BITINODES if we must alter
	 * the allocator to accommodate the request.
	 */
	if ((mp->m_flags & XFS_MOUNT_SMALL_INUMS) && ino > XFS_MAXINUMBER_32)
		mp->m_flags |= XFS_MOUNT_32BITINODES;
	else
		mp->m_flags &= ~XFS_MOUNT_32BITINODES;
E
Eric Sandeen 已提交
648

649
	for (index = 0; index < agcount; index++) {
650
		struct xfs_perag	*pag;
651

652
		ino = XFS_AGINO_TO_INO(mp, index, agino);
653 654 655

		pag = xfs_perag_get(mp, index);

656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
		if (mp->m_flags & XFS_MOUNT_32BITINODES) {
			if (ino > XFS_MAXINUMBER_32) {
				pag->pagi_inodeok = 0;
				pag->pagf_metadata = 0;
			} else {
				pag->pagi_inodeok = 1;
				maxagi++;
				if (index < max_metadata)
					pag->pagf_metadata = 1;
				else
					pag->pagf_metadata = 0;
			}
		} else {
			pag->pagi_inodeok = 1;
			pag->pagf_metadata = 0;
		}
672 673 674 675

		xfs_perag_put(pag);
	}

676
	return (mp->m_flags & XFS_MOUNT_32BITINODES) ? maxagi : agcount;
677 678
}

H
Hannes Eder 已提交
679
STATIC int
L
Linus Torvalds 已提交
680 681 682 683 684 685 686
xfs_blkdev_get(
	xfs_mount_t		*mp,
	const char		*name,
	struct block_device	**bdevp)
{
	int			error = 0;

687 688
	*bdevp = blkdev_get_by_path(name, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
				    mp);
L
Linus Torvalds 已提交
689 690
	if (IS_ERR(*bdevp)) {
		error = PTR_ERR(*bdevp);
691
		xfs_warn(mp, "Invalid device [%s], error=%d", name, error);
L
Linus Torvalds 已提交
692 693
	}

D
Dave Chinner 已提交
694
	return error;
L
Linus Torvalds 已提交
695 696
}

H
Hannes Eder 已提交
697
STATIC void
L
Linus Torvalds 已提交
698 699 700 701
xfs_blkdev_put(
	struct block_device	*bdev)
{
	if (bdev)
702
		blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
L
Linus Torvalds 已提交
703 704
}

705 706 707 708
void
xfs_blkdev_issue_flush(
	xfs_buftarg_t		*buftarg)
{
709
	blkdev_issue_flush(buftarg->bt_bdev, GFP_NOFS, NULL);
710
}
L
Linus Torvalds 已提交
711

712 713 714 715 716
STATIC void
xfs_close_devices(
	struct xfs_mount	*mp)
{
	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
717
		struct block_device *logdev = mp->m_logdev_targp->bt_bdev;
718
		xfs_free_buftarg(mp, mp->m_logdev_targp);
719
		xfs_blkdev_put(logdev);
720 721
	}
	if (mp->m_rtdev_targp) {
722
		struct block_device *rtdev = mp->m_rtdev_targp->bt_bdev;
723
		xfs_free_buftarg(mp, mp->m_rtdev_targp);
724
		xfs_blkdev_put(rtdev);
725
	}
726
	xfs_free_buftarg(mp, mp->m_ddev_targp);
727 728 729 730 731 732 733 734 735 736 737 738 739 740
}

/*
 * 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(
741
	struct xfs_mount	*mp)
742 743 744 745 746 747 748 749
{
	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.
	 */
750 751
	if (mp->m_logname) {
		error = xfs_blkdev_get(mp, mp->m_logname, &logdev);
752 753 754 755
		if (error)
			goto out;
	}

756 757
	if (mp->m_rtname) {
		error = xfs_blkdev_get(mp, mp->m_rtname, &rtdev);
758 759 760 761
		if (error)
			goto out_close_logdev;

		if (rtdev == ddev || rtdev == logdev) {
762 763
			xfs_warn(mp,
	"Cannot mount filesystem with identical rtdev and ddev/logdev.");
D
Dave Chinner 已提交
764
			error = -EINVAL;
765 766 767 768 769 770 771
			goto out_close_rtdev;
		}
	}

	/*
	 * Setup xfs_mount buffer target pointers
	 */
D
Dave Chinner 已提交
772
	error = -ENOMEM;
773
	mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev);
774 775 776 777
	if (!mp->m_ddev_targp)
		goto out_close_rtdev;

	if (rtdev) {
778
		mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev);
779 780 781 782 783
		if (!mp->m_rtdev_targp)
			goto out_free_ddev_targ;
	}

	if (logdev && logdev != ddev) {
784
		mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev);
785 786 787 788 789 790 791 792 793 794
		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)
795
		xfs_free_buftarg(mp, mp->m_rtdev_targp);
796
 out_free_ddev_targ:
797
	xfs_free_buftarg(mp, mp->m_ddev_targp);
798
 out_close_rtdev:
799
	xfs_blkdev_put(rtdev);
800 801 802 803 804 805 806
 out_close_logdev:
	if (logdev && logdev != ddev)
		xfs_blkdev_put(logdev);
 out:
	return error;
}

807 808 809 810 811 812 813 814
/*
 * Setup xfs_mount buffer target pointers based on superblock
 */
STATIC int
xfs_setup_devices(
	struct xfs_mount	*mp)
{
	int			error;
815

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

	return 0;
}
839

840 841 842 843
STATIC int
xfs_init_mount_workqueues(
	struct xfs_mount	*mp)
{
844
	mp->m_buf_workqueue = alloc_workqueue("xfs-buf/%s",
845
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 1, mp->m_fsname);
846 847 848
	if (!mp->m_buf_workqueue)
		goto out;

849
	mp->m_data_workqueue = alloc_workqueue("xfs-data/%s",
850
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0, mp->m_fsname);
851
	if (!mp->m_data_workqueue)
852
		goto out_destroy_buf;
853 854

	mp->m_unwritten_workqueue = alloc_workqueue("xfs-conv/%s",
855
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0, mp->m_fsname);
856 857 858
	if (!mp->m_unwritten_workqueue)
		goto out_destroy_data_iodone_queue;

859
	mp->m_cil_workqueue = alloc_workqueue("xfs-cil/%s",
860
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0, mp->m_fsname);
861 862
	if (!mp->m_cil_workqueue)
		goto out_destroy_unwritten;
D
Dave Chinner 已提交
863 864

	mp->m_reclaim_workqueue = alloc_workqueue("xfs-reclaim/%s",
865
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0, mp->m_fsname);
D
Dave Chinner 已提交
866 867 868 869
	if (!mp->m_reclaim_workqueue)
		goto out_destroy_cil;

	mp->m_log_workqueue = alloc_workqueue("xfs-log/%s",
870 871
			WQ_MEM_RECLAIM|WQ_FREEZABLE|WQ_HIGHPRI, 0,
			mp->m_fsname);
D
Dave Chinner 已提交
872 873 874
	if (!mp->m_log_workqueue)
		goto out_destroy_reclaim;

875
	mp->m_eofblocks_workqueue = alloc_workqueue("xfs-eofblocks/%s",
876
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0, mp->m_fsname);
877 878 879
	if (!mp->m_eofblocks_workqueue)
		goto out_destroy_log;

880 881
	return 0;

882 883
out_destroy_log:
	destroy_workqueue(mp->m_log_workqueue);
D
Dave Chinner 已提交
884 885 886 887
out_destroy_reclaim:
	destroy_workqueue(mp->m_reclaim_workqueue);
out_destroy_cil:
	destroy_workqueue(mp->m_cil_workqueue);
888 889
out_destroy_unwritten:
	destroy_workqueue(mp->m_unwritten_workqueue);
890 891
out_destroy_data_iodone_queue:
	destroy_workqueue(mp->m_data_workqueue);
892 893
out_destroy_buf:
	destroy_workqueue(mp->m_buf_workqueue);
894 895 896 897 898 899 900 901
out:
	return -ENOMEM;
}

STATIC void
xfs_destroy_mount_workqueues(
	struct xfs_mount	*mp)
{
902
	destroy_workqueue(mp->m_eofblocks_workqueue);
D
Dave Chinner 已提交
903 904
	destroy_workqueue(mp->m_log_workqueue);
	destroy_workqueue(mp->m_reclaim_workqueue);
905
	destroy_workqueue(mp->m_cil_workqueue);
906 907
	destroy_workqueue(mp->m_data_workqueue);
	destroy_workqueue(mp->m_unwritten_workqueue);
908
	destroy_workqueue(mp->m_buf_workqueue);
909 910
}

D
Dave Chinner 已提交
911 912 913 914 915 916 917 918 919 920 921 922 923
/*
 * 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)) {
924
		sync_inodes_sb(sb);
D
Dave Chinner 已提交
925 926 927 928
		up_read(&sb->s_umount);
	}
}

929
/* Catch misguided souls that try to use this interface on XFS */
L
Linus Torvalds 已提交
930
STATIC struct inode *
931
xfs_fs_alloc_inode(
L
Linus Torvalds 已提交
932 933
	struct super_block	*sb)
{
934
	BUG();
935
	return NULL;
L
Linus Torvalds 已提交
936 937
}

938
/*
939
 * Now that the generic code is guaranteed not to be accessing
D
Dave Chinner 已提交
940
 * the linux inode, we can inactivate and reclaim the inode.
941
 */
L
Linus Torvalds 已提交
942
STATIC void
943
xfs_fs_destroy_inode(
C
Christoph Hellwig 已提交
944
	struct inode		*inode)
L
Linus Torvalds 已提交
945
{
C
Christoph Hellwig 已提交
946
	struct xfs_inode	*ip = XFS_I(inode);
947
	int			error;
C
Christoph Hellwig 已提交
948

C
Christoph Hellwig 已提交
949
	trace_xfs_destroy_inode(ip);
950

951
	ASSERT(!rwsem_is_locked(&inode->i_rwsem));
D
Dave Chinner 已提交
952 953 954
	XFS_STATS_INC(ip->i_mount, vn_rele);
	XFS_STATS_INC(ip->i_mount, vn_remove);

955 956 957 958 959 960 961
	if (xfs_is_reflink_inode(ip)) {
		error = xfs_reflink_cancel_cow_range(ip, 0, NULLFILEOFF);
		if (error && !XFS_FORCED_SHUTDOWN(ip->i_mount))
			xfs_warn(ip->i_mount,
"Error %d while evicting CoW blocks for inode %llu.",
					error, ip->i_ino);
	}
962

D
Dave Chinner 已提交
963
	xfs_inactive(ip);
C
Christoph Hellwig 已提交
964 965

	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);
D
Dave Chinner 已提交
966
	XFS_STATS_INC(ip->i_mount, vn_reclaim);
C
Christoph Hellwig 已提交
967 968 969 970 971 972 973 974

	/*
	 * 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));

	/*
975 976 977 978 979
	 * 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
980
	 */
981
	xfs_inode_set_reclaim_tag(ip);
L
Linus Torvalds 已提交
982 983
}

984 985 986 987
/*
 * 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
988
 * from the slab. This avoids the need to repeatedly initialise
989 990 991
 * fields in the xfs inode that left in the initialise state
 * when freeing the inode.
 */
992 993
STATIC void
xfs_fs_inode_init_once(
994 995 996 997 998
	void			*inode)
{
	struct xfs_inode	*ip = inode;

	memset(ip, 0, sizeof(struct xfs_inode));
999 1000 1001 1002 1003

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

	/* xfs inode */
1004 1005 1006
	atomic_set(&ip->i_pincount, 0);
	spin_lock_init(&ip->i_flags_lock);

D
Dave Chinner 已提交
1007 1008
	mrlock_init(&ip->i_mmaplock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
		     "xfsino", ip->i_ino);
1009 1010 1011 1012
	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
		     "xfsino", ip->i_ino);
}

1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
/*
 * 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);

1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
	/*
	 * If this unlinked inode is in the middle of recovery, don't
	 * drop the inode just yet; log recovery will take care of
	 * that.  See the comment for this inode flag.
	 */
	if (ip->i_flags & XFS_IRECOVERY) {
		ASSERT(ip->i_mount->m_log->l_flags & XLOG_RECOVERY_NEEDED);
		return 0;
	}

1036 1037 1038
	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
}

1039 1040 1041 1042 1043 1044 1045 1046 1047
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
1048
STATIC int
C
Christoph Hellwig 已提交
1049
xfs_fs_sync_fs(
L
Linus Torvalds 已提交
1050 1051 1052
	struct super_block	*sb,
	int			wait)
{
1053
	struct xfs_mount	*mp = XFS_M(sb);
L
Linus Torvalds 已提交
1054

1055
	/*
C
Christoph Hellwig 已提交
1056
	 * Doing anything during the async pass would be counterproductive.
1057
	 */
C
Christoph Hellwig 已提交
1058
	if (!wait)
C
Christoph Hellwig 已提交
1059 1060
		return 0;

D
Dave Chinner 已提交
1061
	xfs_log_force(mp, XFS_LOG_SYNC);
C
Christoph Hellwig 已提交
1062
	if (laptop_mode) {
L
Linus Torvalds 已提交
1063 1064
		/*
		 * The disk must be active because we're syncing.
1065
		 * We schedule log work now (now that the disk is
L
Linus Torvalds 已提交
1066 1067
		 * active) instead of later (when it might not be).
		 */
1068
		flush_delayed_work(&mp->m_log->l_work);
L
Linus Torvalds 已提交
1069 1070
	}

C
Christoph Hellwig 已提交
1071
	return 0;
L
Linus Torvalds 已提交
1072 1073 1074
}

STATIC int
1075
xfs_fs_statfs(
1076
	struct dentry		*dentry,
L
Linus Torvalds 已提交
1077 1078
	struct kstatfs		*statp)
{
C
Christoph Hellwig 已提交
1079 1080
	struct xfs_mount	*mp = XFS_M(dentry->d_sb);
	xfs_sb_t		*sbp = &mp->m_sb;
1081
	struct xfs_inode	*ip = XFS_I(d_inode(dentry));
C
Christoph Hellwig 已提交
1082
	__uint64_t		fakeinos, id;
1083
	__uint64_t		icount;
1084
	__uint64_t		ifree;
1085
	__uint64_t		fdblocks;
C
Christoph Hellwig 已提交
1086
	xfs_extlen_t		lsize;
1087
	__int64_t		ffree;
C
Christoph Hellwig 已提交
1088 1089 1090 1091 1092 1093 1094 1095

	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);

1096
	icount = percpu_counter_sum(&mp->m_icount);
1097
	ifree = percpu_counter_sum(&mp->m_ifree);
1098
	fdblocks = percpu_counter_sum(&mp->m_fdblocks);
C
Christoph Hellwig 已提交
1099 1100 1101 1102 1103

	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;
1104 1105
	spin_unlock(&mp->m_sb_lock);

1106
	statp->f_bfree = fdblocks - mp->m_alloc_set_aside;
1107 1108
	statp->f_bavail = statp->f_bfree;

C
Christoph Hellwig 已提交
1109
	fakeinos = statp->f_bfree << sbp->sb_inopblog;
1110
	statp->f_files = MIN(icount + fakeinos, (__uint64_t)XFS_MAXINUMBER);
C
Christoph Hellwig 已提交
1111
	if (mp->m_maxicount)
C
Christoph Hellwig 已提交
1112 1113 1114
		statp->f_files = min_t(typeof(statp->f_files),
					statp->f_files,
					mp->m_maxicount);
1115

1116 1117 1118 1119 1120
	/* If sb_icount overshot maxicount, report actual allocation */
	statp->f_files = max_t(typeof(statp->f_files),
					statp->f_files,
					sbp->sb_icount);

1121
	/* make sure statp->f_ffree does not underflow */
1122
	ffree = statp->f_files - (icount - ifree);
1123 1124
	statp->f_ffree = max_t(__int64_t, ffree, 0);

C
Christoph Hellwig 已提交
1125

1126
	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
1127 1128
	    ((mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))) ==
			      (XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))
C
Christoph Hellwig 已提交
1129
		xfs_qm_statvfs(ip, statp);
C
Christoph Hellwig 已提交
1130
	return 0;
L
Linus Torvalds 已提交
1131 1132
}

E
Eric Sandeen 已提交
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
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);
}

1156 1157 1158 1159 1160
/*
 * 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
1161 1162
 * there is no log replay required to write the inodes to disk - this is the
 * primary difference between a sync and a quiesce.
1163
 *
1164 1165
 * Note: xfs_log_quiesce() stops background log work - the callers must ensure
 * it is started again when appropriate.
1166
 */
1167
void
1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
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.");
	/*
1190 1191
	 * Just warn here till VFS can correctly support
	 * read-only remount without racing.
1192
	 */
1193
	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
1194

1195
	xfs_log_quiesce(mp);
1196 1197
}

E
Eric Sandeen 已提交
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218
STATIC int
xfs_test_remount_options(
	struct super_block	*sb,
	struct xfs_mount	*mp,
	char			*options)
{
	int			error = 0;
	struct xfs_mount	*tmp_mp;

	tmp_mp = kmem_zalloc(sizeof(*tmp_mp), KM_MAYFAIL);
	if (!tmp_mp)
		return -ENOMEM;

	tmp_mp->m_super = sb;
	error = xfs_parseargs(tmp_mp, options);
	xfs_free_fsname(tmp_mp);
	kfree(tmp_mp);

	return error;
}

L
Linus Torvalds 已提交
1219
STATIC int
1220
xfs_fs_remount(
L
Linus Torvalds 已提交
1221 1222 1223 1224
	struct super_block	*sb,
	int			*flags,
	char			*options)
{
1225
	struct xfs_mount	*mp = XFS_M(sb);
1226
	xfs_sb_t		*sbp = &mp->m_sb;
1227 1228
	substring_t		args[MAX_OPT_ARGS];
	char			*p;
1229
	int			error;
L
Linus Torvalds 已提交
1230

E
Eric Sandeen 已提交
1231 1232 1233 1234 1235
	/* First, check for complete junk; i.e. invalid options */
	error = xfs_test_remount_options(sb, mp, options);
	if (error)
		return error;

1236
	sync_filesystem(sb);
1237 1238
	while ((p = strsep(&options, ",")) != NULL) {
		int token;
1239

1240 1241
		if (!*p)
			continue;
1242

1243 1244 1245
		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_barrier:
1246
			xfs_warn(mp, "%s option is deprecated, ignoring.", p);
1247
			mp->m_flags |= XFS_MOUNT_BARRIER;
1248 1249
			break;
		case Opt_nobarrier:
1250
			xfs_warn(mp, "%s option is deprecated, ignoring.", p);
1251
			mp->m_flags &= ~XFS_MOUNT_BARRIER;
1252
			break;
1253
		case Opt_inode64:
1254 1255
			mp->m_flags &= ~XFS_MOUNT_SMALL_INUMS;
			mp->m_maxagi = xfs_set_inode_alloc(mp, sbp->sb_agcount);
1256
			break;
1257
		case Opt_inode32:
1258 1259
			mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
			mp->m_maxagi = xfs_set_inode_alloc(mp, sbp->sb_agcount);
1260
			break;
1261
		default:
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
			/*
			 * 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
1279
			xfs_info(mp,
1280
		"mount option \"%s\" not supported for remount", p);
1281
			return -EINVAL;
1282
#else
1283
			break;
1284
#endif
1285
		}
1286 1287
	}

1288
	/* ro -> rw */
1289
	if ((mp->m_flags & XFS_MOUNT_RDONLY) && !(*flags & MS_RDONLY)) {
1290 1291 1292 1293 1294 1295
		if (mp->m_flags & XFS_MOUNT_NORECOVERY) {
			xfs_warn(mp,
		"ro->rw transition prohibited on norecovery mount");
			return -EINVAL;
		}

1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
		if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
		    xfs_sb_has_ro_compat_feature(sbp,
					XFS_SB_FEAT_RO_COMPAT_UNKNOWN)) {
			xfs_warn(mp,
"ro->rw transition prohibited on unknown (0x%x) ro-compat filesystem",
				(sbp->sb_features_ro_compat &
					XFS_SB_FEAT_RO_COMPAT_UNKNOWN));
			return -EINVAL;
		}

1306
		mp->m_flags &= ~XFS_MOUNT_RDONLY;
1307 1308 1309 1310 1311

		/*
		 * If this is the first remount to writeable state we
		 * might have some superblock changes to update.
		 */
1312 1313
		if (mp->m_update_sb) {
			error = xfs_sync_sb(mp, false);
1314
			if (error) {
1315
				xfs_warn(mp, "failed to write sb changes");
1316 1317
				return error;
			}
1318
			mp->m_update_sb = false;
1319
		}
1320 1321 1322 1323 1324

		/*
		 * 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
1325
		xfs_restore_resvblks(mp);
1326
		xfs_log_work_queue(mp);
1327
		xfs_queue_eofblocks(mp);
1328 1329 1330 1331 1332 1333 1334 1335 1336

		/* Recover any CoW blocks that never got remapped. */
		error = xfs_reflink_recover_cow(mp);
		if (error) {
			xfs_err(mp,
	"Error %d recovering leftover CoW allocations.", error);
			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
			return error;
		}
1337 1338 1339 1340 1341

		/* Create the per-AG metadata reservation pool .*/
		error = xfs_fs_reserve_ag_blocks(mp);
		if (error && error != -ENOSPC)
			return error;
1342 1343 1344 1345
	}

	/* rw -> ro */
	if (!(mp->m_flags & XFS_MOUNT_RDONLY) && (*flags & MS_RDONLY)) {
1346 1347 1348 1349 1350 1351 1352
		/* Free the per-AG metadata reservation pool. */
		error = xfs_fs_unreserve_ag_blocks(mp);
		if (error) {
			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
			return error;
		}

1353
		/*
D
Dave Chinner 已提交
1354 1355 1356 1357 1358
		 * 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.
1359
		 */
E
Eric Sandeen 已提交
1360
		xfs_save_resvblks(mp);
1361 1362 1363 1364 1365 1366 1367

		/*
		 * Cancel background eofb scanning so it cannot race with the
		 * final log force+buftarg wait and deadlock the remount.
		 */
		cancel_delayed_work_sync(&mp->m_eofblocks_work);

D
David Chinner 已提交
1368
		xfs_quiesce_attr(mp);
1369 1370 1371
		mp->m_flags |= XFS_MOUNT_RDONLY;
	}

1372
	return 0;
L
Linus Torvalds 已提交
1373 1374
}

C
Christoph Hellwig 已提交
1375 1376
/*
 * Second stage of a freeze. The data is already frozen so we only
1377 1378 1379
 * need to take care of the metadata. Once that's done sync the superblock
 * to the log to dirty it in case of a crash while frozen. This ensures that we
 * will recover the unlinked inode lists on the next mount.
C
Christoph Hellwig 已提交
1380
 */
1381 1382
STATIC int
xfs_fs_freeze(
L
Linus Torvalds 已提交
1383 1384
	struct super_block	*sb)
{
C
Christoph Hellwig 已提交
1385 1386
	struct xfs_mount	*mp = XFS_M(sb);

E
Eric Sandeen 已提交
1387
	xfs_save_resvblks(mp);
D
David Chinner 已提交
1388
	xfs_quiesce_attr(mp);
1389
	return xfs_sync_sb(mp, true);
L
Linus Torvalds 已提交
1390 1391
}

E
Eric Sandeen 已提交
1392 1393 1394 1395 1396 1397 1398
STATIC int
xfs_fs_unfreeze(
	struct super_block	*sb)
{
	struct xfs_mount	*mp = XFS_M(sb);

	xfs_restore_resvblks(mp);
1399
	xfs_log_work_queue(mp);
E
Eric Sandeen 已提交
1400 1401 1402
	return 0;
}

L
Linus Torvalds 已提交
1403
STATIC int
1404
xfs_fs_show_options(
L
Linus Torvalds 已提交
1405
	struct seq_file		*m,
1406
	struct dentry		*root)
L
Linus Torvalds 已提交
1407
{
D
Dave Chinner 已提交
1408
	return xfs_showargs(XFS_M(root->d_sb), m);
L
Linus Torvalds 已提交
1409 1410
}

1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
/*
 * 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);

1421
	/* Fail a mount where the logbuf is smaller than the log stripe */
1422
	if (xfs_sb_version_haslogv2(&mp->m_sb)) {
1423 1424
		if (mp->m_logbsize <= 0 &&
		    mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE) {
1425
			mp->m_logbsize = mp->m_sb.sb_logsunit;
1426 1427
		} else if (mp->m_logbsize > 0 &&
			   mp->m_logbsize < mp->m_sb.sb_logsunit) {
1428 1429
			xfs_warn(mp,
		"logbuf size must be greater than or equal to log stripe size");
D
Dave Chinner 已提交
1430
			return -EINVAL;
1431 1432 1433
		}
	} else {
		/* Fail a mount if the logbuf is larger than 32K */
1434
		if (mp->m_logbsize > XLOG_BIG_RECORD_BSIZE) {
1435 1436
			xfs_warn(mp,
		"logbuf size for version 1 logs must be 16K or 32K");
D
Dave Chinner 已提交
1437
			return -EINVAL;
1438 1439 1440
		}
	}

1441 1442 1443 1444 1445
	/*
	 * V5 filesystems always use attr2 format for attributes.
	 */
	if (xfs_sb_version_hascrc(&mp->m_sb) &&
	    (mp->m_flags & XFS_MOUNT_NOATTR2)) {
1446 1447
		xfs_warn(mp, "Cannot mount a V5 filesystem as noattr2. "
			     "attr2 is always enabled for V5 filesystems.");
D
Dave Chinner 已提交
1448
		return -EINVAL;
1449 1450
	}

1451 1452 1453 1454 1455
	/*
	 * 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) &&
1456
	    !(mp->m_flags & XFS_MOUNT_NOATTR2))
1457 1458 1459 1460 1461 1462
		mp->m_flags |= XFS_MOUNT_ATTR2;

	/*
	 * prohibit r/w mounts of read-only filesystems
	 */
	if ((mp->m_sb.sb_flags & XFS_SBF_READONLY) && !ronly) {
1463 1464
		xfs_warn(mp,
			"cannot mount a read-only filesystem as read-write");
D
Dave Chinner 已提交
1465
		return -EROFS;
1466 1467
	}

1468 1469 1470 1471 1472
	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");
D
Dave Chinner 已提交
1473
		return -EINVAL;
1474 1475
	}

1476 1477 1478
	return 0;
}

D
Dave Chinner 已提交
1479 1480 1481 1482 1483 1484 1485 1486
static int
xfs_init_percpu_counters(
	struct xfs_mount	*mp)
{
	int		error;

	error = percpu_counter_init(&mp->m_icount, 0, GFP_KERNEL);
	if (error)
1487
		return -ENOMEM;
D
Dave Chinner 已提交
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523

	error = percpu_counter_init(&mp->m_ifree, 0, GFP_KERNEL);
	if (error)
		goto free_icount;

	error = percpu_counter_init(&mp->m_fdblocks, 0, GFP_KERNEL);
	if (error)
		goto free_ifree;

	return 0;

free_ifree:
	percpu_counter_destroy(&mp->m_ifree);
free_icount:
	percpu_counter_destroy(&mp->m_icount);
	return -ENOMEM;
}

void
xfs_reinit_percpu_counters(
	struct xfs_mount	*mp)
{
	percpu_counter_set(&mp->m_icount, mp->m_sb.sb_icount);
	percpu_counter_set(&mp->m_ifree, mp->m_sb.sb_ifree);
	percpu_counter_set(&mp->m_fdblocks, mp->m_sb.sb_fdblocks);
}

static void
xfs_destroy_percpu_counters(
	struct xfs_mount	*mp)
{
	percpu_counter_destroy(&mp->m_icount);
	percpu_counter_destroy(&mp->m_ifree);
	percpu_counter_destroy(&mp->m_fdblocks);
}

L
Linus Torvalds 已提交
1524
STATIC int
1525
xfs_fs_fill_super(
L
Linus Torvalds 已提交
1526 1527 1528 1529
	struct super_block	*sb,
	void			*data,
	int			silent)
{
1530
	struct inode		*root;
1531
	struct xfs_mount	*mp = NULL;
D
Dave Chinner 已提交
1532
	int			flags = 0, error = -ENOMEM;
1533

C
Christoph Hellwig 已提交
1534 1535
	mp = kzalloc(sizeof(struct xfs_mount), GFP_KERNEL);
	if (!mp)
1536
		goto out;
L
Linus Torvalds 已提交
1537

C
Christoph Hellwig 已提交
1538 1539 1540
	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
1541
	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
1542
	INIT_DELAYED_WORK(&mp->m_eofblocks_work, xfs_eofblocks_worker);
1543
	INIT_DELAYED_WORK(&mp->m_cowblocks_work, xfs_cowblocks_worker);
D
Dave Chinner 已提交
1544
	mp->m_kobj.kobject.kset = xfs_kset;
1545

C
Christoph Hellwig 已提交
1546 1547
	mp->m_super = sb;
	sb->s_fs_info = mp;
L
Linus Torvalds 已提交
1548

C
Christoph Hellwig 已提交
1549
	error = xfs_parseargs(mp, (char *)data);
1550
	if (error)
1551
		goto out_free_fsname;
L
Linus Torvalds 已提交
1552 1553

	sb_min_blocksize(sb, BBSIZE);
1554
	sb->s_xattr = xfs_xattr_handlers;
1555
	sb->s_export_op = &xfs_export_operations;
1556
#ifdef CONFIG_XFS_QUOTA
1557
	sb->s_qcop = &xfs_quotactl_operations;
J
Jan Kara 已提交
1558
	sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
1559
#endif
1560
	sb->s_op = &xfs_super_operations;
L
Linus Torvalds 已提交
1561

1562
	if (silent)
1563 1564
		flags |= XFS_MFSI_QUIET;

1565
	error = xfs_open_devices(mp);
1566
	if (error)
C
Christoph Hellwig 已提交
1567
		goto out_free_fsname;
1568

D
Dave Chinner 已提交
1569
	error = xfs_init_mount_workqueues(mp);
1570 1571
	if (error)
		goto out_close_devices;
C
Christoph Hellwig 已提交
1572

D
Dave Chinner 已提交
1573
	error = xfs_init_percpu_counters(mp);
1574 1575 1576
	if (error)
		goto out_destroy_workqueues;

1577 1578 1579
	/* Allocate stats memory before we do operations that might use it */
	mp->m_stats.xs_stats = alloc_percpu(struct xfsstats);
	if (!mp->m_stats.xs_stats) {
1580
		error = -ENOMEM;
1581 1582 1583
		goto out_destroy_counters;
	}

1584 1585
	error = xfs_readsb(mp, flags);
	if (error)
1586
		goto out_free_stats;
1587 1588

	error = xfs_finish_flags(mp);
1589
	if (error)
1590
		goto out_free_sb;
1591

1592
	error = xfs_setup_devices(mp);
1593
	if (error)
1594
		goto out_free_sb;
1595 1596 1597

	error = xfs_filestream_mount(mp);
	if (error)
1598
		goto out_free_sb;
1599

1600 1601 1602 1603
	/*
	 * 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
1604 1605 1606
	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
1607
	sb->s_maxbytes = xfs_max_file_offset(sb->s_blocksize_bits);
1608
	sb->s_max_links = XFS_MAXLINK;
L
Linus Torvalds 已提交
1609 1610 1611
	sb->s_time_gran = 1;
	set_posix_acl_flag(sb);

D
Dave Chinner 已提交
1612 1613 1614 1615
	/* 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;

D
Dave Chinner 已提交
1616 1617
	if (mp->m_flags & XFS_MOUNT_DAX) {
		xfs_warn(mp,
1618 1619 1620 1621
		"DAX enabled. Warning: EXPERIMENTAL, use at your own risk");

		error = bdev_dax_supported(sb, sb->s_blocksize);
		if (error) {
D
Dave Chinner 已提交
1622
			xfs_alert(mp,
1623
			"DAX unsupported by block device. Turning off DAX.");
D
Dave Chinner 已提交
1624 1625
			mp->m_flags &= ~XFS_MOUNT_DAX;
		}
1626 1627 1628
		if (xfs_sb_version_hasreflink(&mp->m_sb))
			xfs_alert(mp,
		"DAX and reflink have not been tested together!");
D
Dave Chinner 已提交
1629 1630
	}

1631 1632 1633 1634 1635 1636 1637
	if (xfs_sb_version_hasrmapbt(&mp->m_sb)) {
		if (mp->m_sb.sb_rblocks) {
			xfs_alert(mp,
	"EXPERIMENTAL reverse mapping btree not compatible with realtime device!");
			error = -EINVAL;
			goto out_filestream_unmount;
		}
1638 1639
		xfs_alert(mp,
	"EXPERIMENTAL reverse mapping btree feature enabled. Use at your own risk!");
1640
	}
1641

1642 1643 1644 1645
	if (xfs_sb_version_hasreflink(&mp->m_sb))
		xfs_alert(mp,
	"EXPERIMENTAL reflink feature enabled. Use at your own risk!");

1646
	error = xfs_mountfs(mp);
1647
	if (error)
D
Dave Chinner 已提交
1648
		goto out_filestream_unmount;
1649

1650
	root = igrab(VFS_I(mp->m_rootip));
1651
	if (!root) {
D
Dave Chinner 已提交
1652
		error = -ENOENT;
1653
		goto out_unmount;
C
Christoph Hellwig 已提交
1654
	}
1655
	sb->s_root = d_make_root(root);
1656
	if (!sb->s_root) {
D
Dave Chinner 已提交
1657
		error = -ENOMEM;
1658
		goto out_unmount;
L
Linus Torvalds 已提交
1659
	}
1660

L
Linus Torvalds 已提交
1661
	return 0;
D
Dave Chinner 已提交
1662

D
Dave Chinner 已提交
1663
 out_filestream_unmount:
1664
	xfs_filestream_unmount(mp);
1665 1666
 out_free_sb:
	xfs_freesb(mp);
1667 1668
 out_free_stats:
	free_percpu(mp->m_stats.xs_stats);
1669
 out_destroy_counters:
D
Dave Chinner 已提交
1670
	xfs_destroy_percpu_counters(mp);
1671
 out_destroy_workqueues:
1672
	xfs_destroy_mount_workqueues(mp);
1673
 out_close_devices:
1674
	xfs_close_devices(mp);
1675 1676
 out_free_fsname:
	xfs_free_fsname(mp);
C
Christoph Hellwig 已提交
1677
	kfree(mp);
1678
 out:
D
Dave Chinner 已提交
1679
	return error;
1680

1681
 out_unmount:
1682
	xfs_filestream_unmount(mp);
1683
	xfs_unmountfs(mp);
1684
	goto out_free_sb;
L
Linus Torvalds 已提交
1685 1686
}

D
Dave Chinner 已提交
1687 1688 1689 1690 1691 1692
STATIC void
xfs_fs_put_super(
	struct super_block	*sb)
{
	struct xfs_mount	*mp = XFS_M(sb);

1693
	xfs_notice(mp, "Unmounting Filesystem");
D
Dave Chinner 已提交
1694 1695 1696 1697
	xfs_filestream_unmount(mp);
	xfs_unmountfs(mp);

	xfs_freesb(mp);
1698
	free_percpu(mp->m_stats.xs_stats);
D
Dave Chinner 已提交
1699 1700 1701 1702 1703 1704 1705
	xfs_destroy_percpu_counters(mp);
	xfs_destroy_mount_workqueues(mp);
	xfs_close_devices(mp);
	xfs_free_fsname(mp);
	kfree(mp);
}

A
Al Viro 已提交
1706 1707
STATIC struct dentry *
xfs_fs_mount(
L
Linus Torvalds 已提交
1708 1709 1710
	struct file_system_type	*fs_type,
	int			flags,
	const char		*dev_name,
A
Al Viro 已提交
1711
	void			*data)
L
Linus Torvalds 已提交
1712
{
A
Al Viro 已提交
1713
	return mount_bdev(fs_type, flags, dev_name, data, xfs_fs_fill_super);
1714 1715
}

1716
static long
1717
xfs_fs_nr_cached_objects(
1718
	struct super_block	*sb,
1719
	struct shrink_control	*sc)
1720 1721 1722 1723
{
	return xfs_reclaim_inodes_count(XFS_M(sb));
}

1724
static long
1725 1726
xfs_fs_free_cached_objects(
	struct super_block	*sb,
1727
	struct shrink_control	*sc)
1728
{
1729
	return xfs_reclaim_inodes_nr(XFS_M(sb), sc->nr_to_scan);
1730 1731
}

1732
static const struct super_operations xfs_super_operations = {
1733 1734
	.alloc_inode		= xfs_fs_alloc_inode,
	.destroy_inode		= xfs_fs_destroy_inode,
1735
	.drop_inode		= xfs_fs_drop_inode,
1736
	.put_super		= xfs_fs_put_super,
C
Christoph Hellwig 已提交
1737
	.sync_fs		= xfs_fs_sync_fs,
1738
	.freeze_fs		= xfs_fs_freeze,
E
Eric Sandeen 已提交
1739
	.unfreeze_fs		= xfs_fs_unfreeze,
1740 1741 1742
	.statfs			= xfs_fs_statfs,
	.remount_fs		= xfs_fs_remount,
	.show_options		= xfs_fs_show_options,
1743 1744
	.nr_cached_objects	= xfs_fs_nr_cached_objects,
	.free_cached_objects	= xfs_fs_free_cached_objects,
L
Linus Torvalds 已提交
1745 1746
};

A
Andrew Morton 已提交
1747
static struct file_system_type xfs_fs_type = {
L
Linus Torvalds 已提交
1748 1749
	.owner			= THIS_MODULE,
	.name			= "xfs",
A
Al Viro 已提交
1750
	.mount			= xfs_fs_mount,
L
Linus Torvalds 已提交
1751 1752 1753
	.kill_sb		= kill_block_super,
	.fs_flags		= FS_REQUIRES_DEV,
};
1754
MODULE_ALIAS_FS("xfs");
L
Linus Torvalds 已提交
1755

1756 1757 1758
STATIC int __init
xfs_init_zones(void)
{
1759 1760 1761
	xfs_ioend_bioset = bioset_create(4 * MAX_BUF_PER_PAGE,
			offsetof(struct xfs_ioend, io_inline_bio));
	if (!xfs_ioend_bioset)
1762
		goto out;
1763 1764 1765 1766

	xfs_log_ticket_zone = kmem_zone_init(sizeof(xlog_ticket_t),
						"xfs_log_ticket");
	if (!xfs_log_ticket_zone)
1767
		goto out_free_ioend_bioset;
1768

1769
	xfs_bmap_free_item_zone = kmem_zone_init(
1770
			sizeof(struct xfs_extent_free_item),
1771
			"xfs_bmap_free_item");
1772 1773
	if (!xfs_bmap_free_item_zone)
		goto out_destroy_log_ticket_zone;
1774

1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
	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)
1787
		goto out_destroy_da_state_zone;
1788 1789 1790 1791 1792

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

1793 1794 1795 1796 1797 1798
	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;

1799 1800 1801 1802 1803
	/*
	 * 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.
	 */
1804 1805
	xfs_buf_item_zone = kmem_zone_init(sizeof(struct xfs_buf_log_item),
					   "xfs_buf_item");
1806
	if (!xfs_buf_item_zone)
1807
		goto out_destroy_log_item_desc_zone;
1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822

	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",
1823 1824
			KM_ZONE_HWALIGN | KM_ZONE_RECLAIM | KM_ZONE_SPREAD |
			KM_ZONE_ACCOUNT, xfs_fs_inode_init_once);
1825 1826 1827 1828 1829 1830 1831 1832
	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
1833 1834 1835 1836
	xfs_icreate_zone = kmem_zone_init(sizeof(struct xfs_icreate_item),
					"xfs_icr");
	if (!xfs_icreate_zone)
		goto out_destroy_ili_zone;
1837

1838
	xfs_rud_zone = kmem_zone_init(sizeof(struct xfs_rud_log_item),
1839 1840 1841 1842
			"xfs_rud_item");
	if (!xfs_rud_zone)
		goto out_destroy_icreate_zone;

1843 1844
	xfs_rui_zone = kmem_zone_init(
			xfs_rui_log_item_sizeof(XFS_RUI_MAX_FAST_EXTENTS),
1845 1846 1847 1848
			"xfs_rui_item");
	if (!xfs_rui_zone)
		goto out_destroy_rud_zone;

1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
	xfs_cud_zone = kmem_zone_init(sizeof(struct xfs_cud_log_item),
			"xfs_cud_item");
	if (!xfs_cud_zone)
		goto out_destroy_rui_zone;

	xfs_cui_zone = kmem_zone_init(
			xfs_cui_log_item_sizeof(XFS_CUI_MAX_FAST_EXTENTS),
			"xfs_cui_item");
	if (!xfs_cui_zone)
		goto out_destroy_cud_zone;

1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
	xfs_bud_zone = kmem_zone_init(sizeof(struct xfs_bud_log_item),
			"xfs_bud_item");
	if (!xfs_bud_zone)
		goto out_destroy_cui_zone;

	xfs_bui_zone = kmem_zone_init(
			xfs_bui_log_item_sizeof(XFS_BUI_MAX_FAST_EXTENTS),
			"xfs_bui_item");
	if (!xfs_bui_zone)
		goto out_destroy_bud_zone;

1871 1872
	return 0;

1873 1874 1875 1876
 out_destroy_bud_zone:
	kmem_zone_destroy(xfs_bud_zone);
 out_destroy_cui_zone:
	kmem_zone_destroy(xfs_cui_zone);
1877 1878 1879 1880
 out_destroy_cud_zone:
	kmem_zone_destroy(xfs_cud_zone);
 out_destroy_rui_zone:
	kmem_zone_destroy(xfs_rui_zone);
1881 1882 1883 1884
 out_destroy_rud_zone:
	kmem_zone_destroy(xfs_rud_zone);
 out_destroy_icreate_zone:
	kmem_zone_destroy(xfs_icreate_zone);
D
Dave Chinner 已提交
1885 1886
 out_destroy_ili_zone:
	kmem_zone_destroy(xfs_ili_zone);
1887 1888 1889 1890 1891 1892 1893 1894
 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);
1895 1896
 out_destroy_log_item_desc_zone:
	kmem_zone_destroy(xfs_log_item_desc_zone);
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
 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);
1909 1910
 out_free_ioend_bioset:
	bioset_free(xfs_ioend_bioset);
1911 1912 1913 1914 1915 1916 1917
 out:
	return -ENOMEM;
}

STATIC void
xfs_destroy_zones(void)
{
1918 1919 1920 1921 1922
	/*
	 * Make sure all delayed rcu free are flushed before we
	 * destroy caches.
	 */
	rcu_barrier();
1923 1924
	kmem_zone_destroy(xfs_bui_zone);
	kmem_zone_destroy(xfs_bud_zone);
1925 1926
	kmem_zone_destroy(xfs_cui_zone);
	kmem_zone_destroy(xfs_cud_zone);
1927 1928
	kmem_zone_destroy(xfs_rui_zone);
	kmem_zone_destroy(xfs_rud_zone);
D
Dave Chinner 已提交
1929
	kmem_zone_destroy(xfs_icreate_zone);
1930 1931 1932 1933 1934
	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);
1935
	kmem_zone_destroy(xfs_log_item_desc_zone);
1936 1937 1938 1939 1940 1941
	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);
1942
	bioset_free(xfs_ioend_bioset);
1943
}
L
Linus Torvalds 已提交
1944

1945 1946 1947
STATIC int __init
xfs_init_workqueues(void)
{
1948 1949 1950 1951 1952 1953
	/*
	 * 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.
	 */
1954 1955
	xfs_alloc_wq = alloc_workqueue("xfsalloc",
			WQ_MEM_RECLAIM|WQ_FREEZABLE, 0);
1956
	if (!xfs_alloc_wq)
D
Dave Chinner 已提交
1957
		return -ENOMEM;
1958

1959 1960 1961
	return 0;
}

1962
STATIC void
1963 1964
xfs_destroy_workqueues(void)
{
1965
	destroy_workqueue(xfs_alloc_wq);
1966 1967
}

L
Linus Torvalds 已提交
1968
STATIC int __init
1969
init_xfs_fs(void)
L
Linus Torvalds 已提交
1970 1971 1972
{
	int			error;

1973 1974
	xfs_check_ondisk_structs();

1975 1976
	printk(KERN_INFO XFS_VERSION_STRING " with "
			 XFS_BUILD_OPTIONS " enabled\n");
L
Linus Torvalds 已提交
1977

1978
	xfs_extent_free_init_defer_op();
1979
	xfs_rmap_update_init_defer_op();
1980
	xfs_refcount_update_init_defer_op();
1981
	xfs_bmap_update_init_defer_op();
1982

1983
	xfs_dir_startup();
L
Linus Torvalds 已提交
1984

1985
	error = xfs_init_zones();
1986 1987 1988
	if (error)
		goto out;

1989
	error = xfs_init_workqueues();
1990
	if (error)
C
Christoph Hellwig 已提交
1991
		goto out_destroy_zones;
1992

1993 1994 1995 1996
	error = xfs_mru_cache_init();
	if (error)
		goto out_destroy_wq;

1997
	error = xfs_buf_init();
1998
	if (error)
1999
		goto out_mru_cache_uninit;
2000 2001 2002 2003 2004 2005 2006 2007

	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
2008

B
Brian Foster 已提交
2009 2010 2011
	xfs_kset = kset_create_and_add("xfs", NULL, fs_kobj);
	if (!xfs_kset) {
		error = -ENOMEM;
2012
		goto out_sysctl_unregister;
B
Brian Foster 已提交
2013 2014
	}

2015 2016 2017 2018 2019 2020 2021 2022 2023
	xfsstats.xs_kobj.kobject.kset = xfs_kset;

	xfsstats.xs_stats = alloc_percpu(struct xfsstats);
	if (!xfsstats.xs_stats) {
		error = -ENOMEM;
		goto out_kset_unregister;
	}

	error = xfs_sysfs_init(&xfsstats.xs_kobj, &xfs_stats_ktype, NULL,
2024 2025
			       "stats");
	if (error)
2026
		goto out_free_stats;
2027

2028 2029 2030
#ifdef DEBUG
	xfs_dbg_kobj.kobject.kset = xfs_kset;
	error = xfs_sysfs_init(&xfs_dbg_kobj, &xfs_dbg_ktype, NULL, "debug");
2031
	if (error)
2032
		goto out_remove_stats_kobj;
2033 2034 2035 2036
#endif

	error = xfs_qm_init();
	if (error)
2037
		goto out_remove_dbg_kobj;
L
Linus Torvalds 已提交
2038 2039 2040

	error = register_filesystem(&xfs_fs_type);
	if (error)
2041
		goto out_qm_exit;
L
Linus Torvalds 已提交
2042 2043
	return 0;

2044 2045
 out_qm_exit:
	xfs_qm_exit();
2046
 out_remove_dbg_kobj:
2047 2048
#ifdef DEBUG
	xfs_sysfs_del(&xfs_dbg_kobj);
2049
 out_remove_stats_kobj:
2050
#endif
2051 2052 2053
	xfs_sysfs_del(&xfsstats.xs_kobj);
 out_free_stats:
	free_percpu(xfsstats.xs_stats);
2054
 out_kset_unregister:
B
Brian Foster 已提交
2055
	kset_unregister(xfs_kset);
2056 2057 2058 2059 2060
 out_sysctl_unregister:
	xfs_sysctl_unregister();
 out_cleanup_procfs:
	xfs_cleanup_procfs();
 out_buf_terminate:
2061
	xfs_buf_terminate();
2062 2063
 out_mru_cache_uninit:
	xfs_mru_cache_uninit();
2064 2065
 out_destroy_wq:
	xfs_destroy_workqueues();
2066
 out_destroy_zones:
2067
	xfs_destroy_zones();
2068
 out:
L
Linus Torvalds 已提交
2069 2070 2071 2072
	return error;
}

STATIC void __exit
2073
exit_xfs_fs(void)
L
Linus Torvalds 已提交
2074
{
2075
	xfs_qm_exit();
L
Linus Torvalds 已提交
2076
	unregister_filesystem(&xfs_fs_type);
2077 2078 2079
#ifdef DEBUG
	xfs_sysfs_del(&xfs_dbg_kobj);
#endif
2080 2081
	xfs_sysfs_del(&xfsstats.xs_kobj);
	free_percpu(xfsstats.xs_stats);
B
Brian Foster 已提交
2082
	kset_unregister(xfs_kset);
2083 2084
	xfs_sysctl_unregister();
	xfs_cleanup_procfs();
2085
	xfs_buf_terminate();
2086
	xfs_mru_cache_uninit();
2087
	xfs_destroy_workqueues();
2088
	xfs_destroy_zones();
2089
	xfs_uuid_table_free();
L
Linus Torvalds 已提交
2090 2091 2092 2093 2094 2095 2096 2097
}

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");