xfs_iops.c 31.8 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2 3
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * 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 18 19
 */
#include "xfs.h"
#include "xfs_fs.h"
20
#include "xfs_shared.h"
21 22 23
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
24
#include "xfs_mount.h"
25
#include "xfs_da_format.h"
L
Linus Torvalds 已提交
26 27
#include "xfs_inode.h"
#include "xfs_bmap.h"
D
Dave Chinner 已提交
28
#include "xfs_bmap_util.h"
29 30
#include "xfs_acl.h"
#include "xfs_quota.h"
L
Linus Torvalds 已提交
31 32
#include "xfs_error.h"
#include "xfs_attr.h"
33
#include "xfs_trans.h"
C
Christoph Hellwig 已提交
34
#include "xfs_trace.h"
35
#include "xfs_icache.h"
D
Dave Chinner 已提交
36
#include "xfs_symlink.h"
37
#include "xfs_da_btree.h"
38
#include "xfs_dir2.h"
Z
Zhi Yong Wu 已提交
39
#include "xfs_trans_space.h"
40
#include "xfs_pnfs.h"
41
#include "xfs_iomap.h"
L
Linus Torvalds 已提交
42

43
#include <linux/capability.h>
L
Linus Torvalds 已提交
44
#include <linux/xattr.h>
45
#include <linux/posix_acl.h>
46
#include <linux/security.h>
47
#include <linux/iomap.h>
48
#include <linux/slab.h>
L
Linus Torvalds 已提交
49

50 51 52 53 54 55 56 57 58 59 60 61
/*
 * Directories have different lock order w.r.t. mmap_sem compared to regular
 * files. This is due to readdir potentially triggering page faults on a user
 * buffer inside filldir(), and this happens with the ilock on the directory
 * held. For regular files, the lock order is the other way around - the
 * mmap_sem is taken during the page fault, and then we lock the ilock to do
 * block mapping. Hence we need a different class for the directory ilock so
 * that lockdep can tell them apart.
 */
static struct lock_class_key xfs_nondir_ilock_class;
static struct lock_class_key xfs_dir_ilock_class;

62 63 64 65 66
static int
xfs_initxattrs(
	struct inode		*inode,
	const struct xattr	*xattr_array,
	void			*fs_info)
67
{
68 69 70
	const struct xattr	*xattr;
	struct xfs_inode	*ip = XFS_I(inode);
	int			error = 0;
71 72

	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
D
Dave Chinner 已提交
73
		error = xfs_attr_set(ip, xattr->name, xattr->value,
74
				      xattr->value_len, ATTR_SECURE);
75 76 77 78 79 80
		if (error < 0)
			break;
	}
	return error;
}

81 82 83 84 85 86
/*
 * Hook in SELinux.  This is not quite correct yet, what we really need
 * here (as we do for default ACLs) is a mechanism by which creation of
 * these attrs can be journalled at inode creation time (along with the
 * inode, of course, such that log replay can't cause these to be lost).
 */
87

88
STATIC int
89
xfs_init_security(
90
	struct inode	*inode,
91 92
	struct inode	*dir,
	const struct qstr *qstr)
93
{
D
Dave Chinner 已提交
94
	return security_inode_init_security(inode, dir, qstr,
95
					     &xfs_initxattrs, NULL);
96 97
}

98 99
static void
xfs_dentry_to_name(
100 101 102 103 104 105 106 107 108 109
	struct xfs_name	*namep,
	struct dentry	*dentry)
{
	namep->name = dentry->d_name.name;
	namep->len = dentry->d_name.len;
	namep->type = XFS_DIR3_FT_UNKNOWN;
}

static int
xfs_dentry_mode_to_name(
110
	struct xfs_name	*namep,
111 112
	struct dentry	*dentry,
	int		mode)
113 114 115
{
	namep->name = dentry->d_name.name;
	namep->len = dentry->d_name.len;
116
	namep->type = xfs_mode_to_ftype(mode);
117 118 119 120 121

	if (unlikely(namep->type == XFS_DIR3_FT_UNKNOWN))
		return -EFSCORRUPTED;

	return 0;
122 123
}

124
STATIC void
125
xfs_cleanup_inode(
126
	struct inode	*dir,
127
	struct inode	*inode,
128
	struct dentry	*dentry)
129
{
130
	struct xfs_name	teardown;
131 132

	/* Oh, the horror.
133
	 * If we can't add the ACL or we fail in
134
	 * xfs_init_security we must back out.
135 136
	 * ENOSPC can hit here, among other things.
	 */
137
	xfs_dentry_to_name(&teardown, dentry);
138

139
	xfs_remove(XFS_I(dir), &teardown, XFS_I(inode));
140 141
}

L
Linus Torvalds 已提交
142
STATIC int
143
xfs_generic_create(
L
Linus Torvalds 已提交
144 145
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
146
	umode_t		mode,
147 148
	dev_t		rdev,
	bool		tmpfile)	/* unnamed file */
L
Linus Torvalds 已提交
149
{
C
Christoph Hellwig 已提交
150
	struct inode	*inode;
151
	struct xfs_inode *ip = NULL;
152
	struct posix_acl *default_acl, *acl;
153
	struct xfs_name	name;
L
Linus Torvalds 已提交
154 155 156 157 158 159
	int		error;

	/*
	 * Irix uses Missed'em'V split, but doesn't want to see
	 * the upper 5 bits of (14bit) major.
	 */
160 161 162 163 164 165
	if (S_ISCHR(mode) || S_ISBLK(mode)) {
		if (unlikely(!sysv_valid_dev(rdev) || MAJOR(rdev) & ~0x1ff))
			return -EINVAL;
	} else {
		rdev = 0;
	}
L
Linus Torvalds 已提交
166

167 168 169
	error = posix_acl_create(dir, &mode, &default_acl, &acl);
	if (error)
		return error;
L
Linus Torvalds 已提交
170

171 172 173 174 175
	/* Verify mode is valid also for tmpfile case */
	error = xfs_dentry_mode_to_name(&name, dentry, mode);
	if (unlikely(error))
		goto out_free_acl;

176 177 178 179 180
	if (!tmpfile) {
		error = xfs_create(XFS_I(dir), &name, mode, rdev, &ip);
	} else {
		error = xfs_create_tmpfile(XFS_I(dir), dentry, mode, &ip);
	}
C
Christoph Hellwig 已提交
181 182
	if (unlikely(error))
		goto out_free_acl;
183

184
	inode = VFS_I(ip);
185

186
	error = xfs_init_security(inode, dir, &dentry->d_name);
C
Christoph Hellwig 已提交
187 188 189
	if (unlikely(error))
		goto out_cleanup_inode;

190
#ifdef CONFIG_XFS_POSIX_ACL
C
Christoph Hellwig 已提交
191
	if (default_acl) {
192
		error = __xfs_set_acl(inode, default_acl, ACL_TYPE_DEFAULT);
193
		if (error)
C
Christoph Hellwig 已提交
194
			goto out_cleanup_inode;
L
Linus Torvalds 已提交
195
	}
196
	if (acl) {
197
		error = __xfs_set_acl(inode, acl, ACL_TYPE_ACCESS);
198 199 200 201
		if (error)
			goto out_cleanup_inode;
	}
#endif
L
Linus Torvalds 已提交
202

203 204
	xfs_setup_iops(ip);

205 206 207 208 209
	if (tmpfile)
		d_tmpfile(dentry, inode);
	else
		d_instantiate(dentry, inode);

210 211
	xfs_finish_inode_setup(ip);

212 213 214 215 216
 out_free_acl:
	if (default_acl)
		posix_acl_release(default_acl);
	if (acl)
		posix_acl_release(acl);
D
Dave Chinner 已提交
217
	return error;
C
Christoph Hellwig 已提交
218 219

 out_cleanup_inode:
220
	xfs_finish_inode_setup(ip);
221 222 223
	if (!tmpfile)
		xfs_cleanup_inode(dir, inode, dentry);
	iput(inode);
224
	goto out_free_acl;
L
Linus Torvalds 已提交
225 226
}

227 228 229 230 231 232 233 234 235 236
STATIC int
xfs_vn_mknod(
	struct inode	*dir,
	struct dentry	*dentry,
	umode_t		mode,
	dev_t		rdev)
{
	return xfs_generic_create(dir, dentry, mode, rdev, false);
}

L
Linus Torvalds 已提交
237
STATIC int
238
xfs_vn_create(
L
Linus Torvalds 已提交
239 240
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
241
	umode_t		mode,
A
Al Viro 已提交
242
	bool		flags)
L
Linus Torvalds 已提交
243
{
244
	return xfs_vn_mknod(dir, dentry, mode, 0);
L
Linus Torvalds 已提交
245 246 247
}

STATIC int
248
xfs_vn_mkdir(
L
Linus Torvalds 已提交
249 250
	struct inode	*dir,
	struct dentry	*dentry,
251
	umode_t		mode)
L
Linus Torvalds 已提交
252
{
253
	return xfs_vn_mknod(dir, dentry, mode|S_IFDIR, 0);
L
Linus Torvalds 已提交
254 255 256
}

STATIC struct dentry *
257
xfs_vn_lookup(
L
Linus Torvalds 已提交
258 259
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
260
	unsigned int flags)
L
Linus Torvalds 已提交
261
{
262
	struct xfs_inode *cip;
263
	struct xfs_name	name;
L
Linus Torvalds 已提交
264 265 266 267 268
	int		error;

	if (dentry->d_name.len >= MAXNAMELEN)
		return ERR_PTR(-ENAMETOOLONG);

269
	xfs_dentry_to_name(&name, dentry);
270
	error = xfs_lookup(XFS_I(dir), &name, &cip, NULL);
271
	if (unlikely(error)) {
D
Dave Chinner 已提交
272 273
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
L
Linus Torvalds 已提交
274 275 276 277
		d_add(dentry, NULL);
		return NULL;
	}

278
	return d_splice_alias(VFS_I(cip), dentry);
L
Linus Torvalds 已提交
279 280
}

281 282 283 284
STATIC struct dentry *
xfs_vn_ci_lookup(
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
285
	unsigned int flags)
286 287 288 289 290 291 292 293 294 295
{
	struct xfs_inode *ip;
	struct xfs_name	xname;
	struct xfs_name ci_name;
	struct qstr	dname;
	int		error;

	if (dentry->d_name.len >= MAXNAMELEN)
		return ERR_PTR(-ENAMETOOLONG);

296
	xfs_dentry_to_name(&xname, dentry);
297 298
	error = xfs_lookup(XFS_I(dir), &xname, &ip, &ci_name);
	if (unlikely(error)) {
D
Dave Chinner 已提交
299 300
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
301 302 303 304 305
		/*
		 * call d_add(dentry, NULL) here when d_drop_negative_children
		 * is called in xfs_vn_mknod (ie. allow negative dentries
		 * with CI filesystems).
		 */
306 307 308 309 310
		return NULL;
	}

	/* if exact match, just splice and exit */
	if (!ci_name.name)
311
		return d_splice_alias(VFS_I(ip), dentry);
312 313 314 315

	/* else case-insensitive match... */
	dname.name = ci_name.name;
	dname.len = ci_name.len;
316
	dentry = d_add_ci(dentry, VFS_I(ip), &dname);
317 318 319 320
	kmem_free(ci_name.name);
	return dentry;
}

L
Linus Torvalds 已提交
321
STATIC int
322
xfs_vn_link(
L
Linus Torvalds 已提交
323 324 325 326
	struct dentry	*old_dentry,
	struct inode	*dir,
	struct dentry	*dentry)
{
327
	struct inode	*inode = d_inode(old_dentry);
328
	struct xfs_name	name;
L
Linus Torvalds 已提交
329 330
	int		error;

331 332 333
	error = xfs_dentry_mode_to_name(&name, dentry, inode->i_mode);
	if (unlikely(error))
		return error;
L
Linus Torvalds 已提交
334

335
	error = xfs_link(XFS_I(dir), XFS_I(inode), &name);
336
	if (unlikely(error))
D
Dave Chinner 已提交
337
		return error;
338

A
Al Viro 已提交
339
	ihold(inode);
340 341
	d_instantiate(dentry, inode);
	return 0;
L
Linus Torvalds 已提交
342 343 344
}

STATIC int
345
xfs_vn_unlink(
L
Linus Torvalds 已提交
346 347 348
	struct inode	*dir,
	struct dentry	*dentry)
{
349
	struct xfs_name	name;
L
Linus Torvalds 已提交
350 351
	int		error;

352
	xfs_dentry_to_name(&name, dentry);
L
Linus Torvalds 已提交
353

354
	error = xfs_remove(XFS_I(dir), &name, XFS_I(d_inode(dentry)));
355 356 357 358 359 360 361 362 363 364 365
	if (error)
		return error;

	/*
	 * With unlink, the VFS makes the dentry "negative": no inode,
	 * but still hashed. This is incompatible with case-insensitive
	 * mode, so invalidate (unhash) the dentry in CI-mode.
	 */
	if (xfs_sb_version_hasasciici(&XFS_M(dir->i_sb)->m_sb))
		d_invalidate(dentry);
	return 0;
L
Linus Torvalds 已提交
366 367 368
}

STATIC int
369
xfs_vn_symlink(
L
Linus Torvalds 已提交
370 371 372 373
	struct inode	*dir,
	struct dentry	*dentry,
	const char	*symname)
{
374 375
	struct inode	*inode;
	struct xfs_inode *cip = NULL;
376
	struct xfs_name	name;
L
Linus Torvalds 已提交
377
	int		error;
A
Al Viro 已提交
378
	umode_t		mode;
L
Linus Torvalds 已提交
379

380
	mode = S_IFLNK |
A
Al Viro 已提交
381
		(irix_symlink_mode ? 0777 & ~current_umask() : S_IRWXUGO);
382 383 384
	error = xfs_dentry_mode_to_name(&name, dentry, mode);
	if (unlikely(error))
		goto out;
L
Linus Torvalds 已提交
385

C
Christoph Hellwig 已提交
386
	error = xfs_symlink(XFS_I(dir), &name, symname, mode, &cip);
387 388 389
	if (unlikely(error))
		goto out;

390
	inode = VFS_I(cip);
391

392
	error = xfs_init_security(inode, dir, &dentry->d_name);
393 394 395
	if (unlikely(error))
		goto out_cleanup_inode;

396 397
	xfs_setup_iops(cip);

398
	d_instantiate(dentry, inode);
399
	xfs_finish_inode_setup(cip);
400 401 402
	return 0;

 out_cleanup_inode:
403
	xfs_finish_inode_setup(cip);
404
	xfs_cleanup_inode(dir, inode, dentry);
405
	iput(inode);
406
 out:
D
Dave Chinner 已提交
407
	return error;
L
Linus Torvalds 已提交
408 409 410
}

STATIC int
411
xfs_vn_rename(
L
Linus Torvalds 已提交
412 413 414
	struct inode	*odir,
	struct dentry	*odentry,
	struct inode	*ndir,
415 416
	struct dentry	*ndentry,
	unsigned int	flags)
L
Linus Torvalds 已提交
417
{
418
	struct inode	*new_inode = d_inode(ndentry);
419
	int		omode = 0;
420
	int		error;
421 422
	struct xfs_name	oname;
	struct xfs_name	nname;
L
Linus Torvalds 已提交
423

D
Dave Chinner 已提交
424
	if (flags & ~(RENAME_NOREPLACE | RENAME_EXCHANGE | RENAME_WHITEOUT))
425 426
		return -EINVAL;

427 428
	/* if we are exchanging files, we need to set i_mode of both files */
	if (flags & RENAME_EXCHANGE)
429
		omode = d_inode(ndentry)->i_mode;
430

431 432 433 434 435 436 437 438
	error = xfs_dentry_mode_to_name(&oname, odentry, omode);
	if (omode && unlikely(error))
		return error;

	error = xfs_dentry_mode_to_name(&nname, ndentry,
					d_inode(odentry)->i_mode);
	if (unlikely(error))
		return error;
439

440
	return xfs_rename(XFS_I(odir), &oname, XFS_I(d_inode(odentry)),
441
			  XFS_I(ndir), &nname,
442
			  new_inode ? XFS_I(new_inode) : NULL, flags);
L
Linus Torvalds 已提交
443 444 445 446 447 448 449
}

/*
 * careful here - this function can get called recursively, so
 * we need to be very careful about how much stack we use.
 * uio is kmalloced for this reason...
 */
450
STATIC const char *
451
xfs_vn_get_link(
L
Linus Torvalds 已提交
452
	struct dentry		*dentry,
453
	struct inode		*inode,
454
	struct delayed_call	*done)
L
Linus Torvalds 已提交
455 456
{
	char			*link;
457
	int			error = -ENOMEM;
L
Linus Torvalds 已提交
458

459 460 461
	if (!dentry)
		return ERR_PTR(-ECHILD);

462
	link = kmalloc(XFS_SYMLINK_MAXLEN+1, GFP_KERNEL);
463 464
	if (!link)
		goto out_err;
L
Linus Torvalds 已提交
465

466
	error = xfs_readlink(XFS_I(d_inode(dentry)), link);
467 468
	if (unlikely(error))
		goto out_kfree;
L
Linus Torvalds 已提交
469

470 471
	set_delayed_call(done, kfree_link, link);
	return link;
472 473 474 475

 out_kfree:
	kfree(link);
 out_err:
476
	return ERR_PTR(error);
L
Linus Torvalds 已提交
477 478
}

479 480 481 482 483 484 485 486 487 488
STATIC const char *
xfs_vn_get_link_inline(
	struct dentry		*dentry,
	struct inode		*inode,
	struct delayed_call	*done)
{
	ASSERT(XFS_I(inode)->i_df.if_flags & XFS_IFINLINE);
	return XFS_I(inode)->i_df.if_u1.if_data;
}

L
Linus Torvalds 已提交
489
STATIC int
490
xfs_vn_getattr(
491 492 493 494
	const struct path	*path,
	struct kstat		*stat,
	u32			request_mask,
	unsigned int		query_flags)
L
Linus Torvalds 已提交
495
{
496
	struct inode		*inode = d_inode(path->dentry);
497 498 499
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;

C
Christoph Hellwig 已提交
500
	trace_xfs_getattr(ip);
501 502

	if (XFS_FORCED_SHUTDOWN(mp))
E
Eric Sandeen 已提交
503
		return -EIO;
504 505 506

	stat->size = XFS_ISIZE(ip);
	stat->dev = inode->i_sb->s_dev;
D
Dave Chinner 已提交
507
	stat->mode = inode->i_mode;
508
	stat->nlink = inode->i_nlink;
509 510
	stat->uid = inode->i_uid;
	stat->gid = inode->i_gid;
511 512
	stat->ino = ip->i_ino;
	stat->atime = inode->i_atime;
513 514
	stat->mtime = inode->i_mtime;
	stat->ctime = inode->i_ctime;
515 516 517
	stat->blocks =
		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);

518 519 520 521 522 523 524 525 526 527 528 529 530 531
	if (ip->i_d.di_version == 3) {
		if (request_mask & STATX_BTIME) {
			stat->result_mask |= STATX_BTIME;
			stat->btime.tv_sec = ip->i_d.di_crtime.t_sec;
			stat->btime.tv_nsec = ip->i_d.di_crtime.t_nsec;
		}
	}

	if (ip->i_d.di_flags & XFS_DIFLAG_IMMUTABLE)
		stat->attributes |= STATX_ATTR_IMMUTABLE;
	if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
		stat->attributes |= STATX_ATTR_APPEND;
	if (ip->i_d.di_flags & XFS_DIFLAG_NODUMP)
		stat->attributes |= STATX_ATTR_NODUMP;
532 533 534 535 536

	switch (inode->i_mode & S_IFMT) {
	case S_IFBLK:
	case S_IFCHR:
		stat->blksize = BLKDEV_IOSIZE;
C
Christoph Hellwig 已提交
537
		stat->rdev = inode->i_rdev;
538 539
		break;
	default:
540
		if (XFS_IS_REALTIME_INODE(ip)) {
541 542 543 544 545 546 547 548 549 550 551
			/*
			 * If the file blocks are being allocated from a
			 * realtime volume, then return the inode's realtime
			 * extent size or the realtime volume's extent size.
			 */
			stat->blksize =
				xfs_get_extsz_hint(ip) << mp->m_sb.sb_blocklog;
		} else
			stat->blksize = xfs_preferred_iosize(mp);
		stat->rdev = 0;
		break;
552
	}
553 554

	return 0;
L
Linus Torvalds 已提交
555 556
}

557 558 559 560 561
static void
xfs_setattr_mode(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
562 563
	struct inode		*inode = VFS_I(ip);
	umode_t			mode = iattr->ia_mode;
564 565 566 567 568 569 570

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

	inode->i_mode &= S_IFMT;
	inode->i_mode |= mode & ~S_IFMT;
}

571
void
C
Christoph Hellwig 已提交
572 573 574 575 576 577 578 579
xfs_setattr_time(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
	struct inode		*inode = VFS_I(ip);

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

580
	if (iattr->ia_valid & ATTR_ATIME)
C
Christoph Hellwig 已提交
581
		inode->i_atime = iattr->ia_atime;
582
	if (iattr->ia_valid & ATTR_CTIME)
C
Christoph Hellwig 已提交
583
		inode->i_ctime = iattr->ia_ctime;
584
	if (iattr->ia_valid & ATTR_MTIME)
C
Christoph Hellwig 已提交
585 586 587
		inode->i_mtime = iattr->ia_mtime;
}

588 589 590 591 592
static int
xfs_vn_change_ok(
	struct dentry	*dentry,
	struct iattr	*iattr)
{
593
	struct xfs_mount	*mp = XFS_I(d_inode(dentry))->i_mount;
594 595 596 597 598 599 600

	if (mp->m_flags & XFS_MOUNT_RDONLY)
		return -EROFS;

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

601
	return setattr_prepare(dentry, iattr);
602 603 604 605 606 607
}

/*
 * Set non-size attributes of an inode.
 *
 * Caution: The caller of this function is responsible for calling
608
 * setattr_prepare() or otherwise verifying the change is fine.
609
 */
C
Christoph Hellwig 已提交
610 611 612 613 614 615 616 617 618 619 620
int
xfs_setattr_nonsize(
	struct xfs_inode	*ip,
	struct iattr		*iattr,
	int			flags)
{
	xfs_mount_t		*mp = ip->i_mount;
	struct inode		*inode = VFS_I(ip);
	int			mask = iattr->ia_valid;
	xfs_trans_t		*tp;
	int			error;
621 622
	kuid_t			uid = GLOBAL_ROOT_UID, iuid = GLOBAL_ROOT_UID;
	kgid_t			gid = GLOBAL_ROOT_GID, igid = GLOBAL_ROOT_GID;
C
Christoph Hellwig 已提交
623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
	struct xfs_dquot	*udqp = NULL, *gdqp = NULL;
	struct xfs_dquot	*olddquot1 = NULL, *olddquot2 = NULL;

	ASSERT((mask & ATTR_SIZE) == 0);

	/*
	 * If disk quotas is on, we make sure that the dquots do exist on disk,
	 * before we start any other transactions. Trying to do this later
	 * is messy. We don't care to take a readlock to look at the ids
	 * in inode here, because we can't hold it across the trans_reserve.
	 * If the IDs do change before we take the ilock, we're covered
	 * because the i_*dquot fields will get updated anyway.
	 */
	if (XFS_IS_QUOTA_ON(mp) && (mask & (ATTR_UID|ATTR_GID))) {
		uint	qflags = 0;

		if ((mask & ATTR_UID) && XFS_IS_UQUOTA_ON(mp)) {
			uid = iattr->ia_uid;
			qflags |= XFS_QMOPT_UQUOTA;
		} else {
643
			uid = inode->i_uid;
C
Christoph Hellwig 已提交
644 645 646 647 648
		}
		if ((mask & ATTR_GID) && XFS_IS_GQUOTA_ON(mp)) {
			gid = iattr->ia_gid;
			qflags |= XFS_QMOPT_GQUOTA;
		}  else {
649
			gid = inode->i_gid;
C
Christoph Hellwig 已提交
650 651 652 653 654 655 656 657 658
		}

		/*
		 * We take a reference when we initialize udqp and gdqp,
		 * so it is important that we never blindly double trip on
		 * the same variable. See xfs_create() for an example.
		 */
		ASSERT(udqp == NULL);
		ASSERT(gdqp == NULL);
659 660 661 662
		error = xfs_qm_vop_dqalloc(ip, xfs_kuid_to_uid(uid),
					   xfs_kgid_to_gid(gid),
					   xfs_get_projid(ip),
					   qflags, &udqp, &gdqp, NULL);
C
Christoph Hellwig 已提交
663 664 665 666
		if (error)
			return error;
	}

667
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ichange, 0, 0, 0, &tp);
C
Christoph Hellwig 已提交
668
	if (error)
669
		goto out_dqrele;
C
Christoph Hellwig 已提交
670 671

	xfs_ilock(ip, XFS_ILOCK_EXCL);
672
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
673 674 675 676 677 678 679 680 681 682 683

	/*
	 * Change file ownership.  Must be the owner or privileged.
	 */
	if (mask & (ATTR_UID|ATTR_GID)) {
		/*
		 * These IDs could have changed since we last looked at them.
		 * But, we're assured that if the ownership did change
		 * while we didn't have the inode locked, inode's dquot(s)
		 * would have changed also.
		 */
684 685
		iuid = inode->i_uid;
		igid = inode->i_gid;
C
Christoph Hellwig 已提交
686 687 688 689 690 691 692 693
		gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
		uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;

		/*
		 * Do a quota reservation only if uid/gid is actually
		 * going to change.
		 */
		if (XFS_IS_QUOTA_RUNNING(mp) &&
694 695
		    ((XFS_IS_UQUOTA_ON(mp) && !uid_eq(iuid, uid)) ||
		     (XFS_IS_GQUOTA_ON(mp) && !gid_eq(igid, gid)))) {
C
Christoph Hellwig 已提交
696 697
			ASSERT(tp);
			error = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
698
						NULL, capable(CAP_FOWNER) ?
C
Christoph Hellwig 已提交
699 700
						XFS_QMOPT_FORCE_RES : 0);
			if (error)	/* out of quota */
701
				goto out_cancel;
C
Christoph Hellwig 已提交
702 703 704 705 706 707 708 709 710 711 712 713 714
		}
	}

	/*
	 * Change file ownership.  Must be the owner or privileged.
	 */
	if (mask & (ATTR_UID|ATTR_GID)) {
		/*
		 * CAP_FSETID overrides the following restrictions:
		 *
		 * The set-user-ID and set-group-ID bits of a file will be
		 * cleared upon successful return from chown()
		 */
D
Dave Chinner 已提交
715
		if ((inode->i_mode & (S_ISUID|S_ISGID)) &&
C
Christoph Hellwig 已提交
716
		    !capable(CAP_FSETID))
D
Dave Chinner 已提交
717
			inode->i_mode &= ~(S_ISUID|S_ISGID);
C
Christoph Hellwig 已提交
718 719 720 721 722

		/*
		 * Change the ownerships and register quota modifications
		 * in the transaction.
		 */
723
		if (!uid_eq(iuid, uid)) {
C
Christoph Hellwig 已提交
724 725 726 727 728 729
			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_UQUOTA_ON(mp)) {
				ASSERT(mask & ATTR_UID);
				ASSERT(udqp);
				olddquot1 = xfs_qm_vop_chown(tp, ip,
							&ip->i_udquot, udqp);
			}
730
			ip->i_d.di_uid = xfs_kuid_to_uid(uid);
C
Christoph Hellwig 已提交
731 732
			inode->i_uid = uid;
		}
733
		if (!gid_eq(igid, gid)) {
C
Christoph Hellwig 已提交
734
			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_GQUOTA_ON(mp)) {
735 736
				ASSERT(xfs_sb_version_has_pquotino(&mp->m_sb) ||
				       !XFS_IS_PQUOTA_ON(mp));
C
Christoph Hellwig 已提交
737 738 739 740 741
				ASSERT(mask & ATTR_GID);
				ASSERT(gdqp);
				olddquot2 = xfs_qm_vop_chown(tp, ip,
							&ip->i_gdquot, gdqp);
			}
742
			ip->i_d.di_gid = xfs_kgid_to_gid(gid);
C
Christoph Hellwig 已提交
743 744 745 746
			inode->i_gid = gid;
		}
	}

747
	if (mask & ATTR_MODE)
748
		xfs_setattr_mode(ip, iattr);
C
Christoph Hellwig 已提交
749 750
	if (mask & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
751 752 753

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

754
	XFS_STATS_INC(mp, xs_ig_attrchg);
C
Christoph Hellwig 已提交
755 756 757

	if (mp->m_flags & XFS_MOUNT_WSYNC)
		xfs_trans_set_sync(tp);
758
	error = xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
759 760 761 762 763 764 765 766 767 768 769 770

	xfs_iunlock(ip, XFS_ILOCK_EXCL);

	/*
	 * Release any dquot(s) the inode had kept before chown.
	 */
	xfs_qm_dqrele(olddquot1);
	xfs_qm_dqrele(olddquot2);
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);

	if (error)
E
Eric Sandeen 已提交
771
		return error;
C
Christoph Hellwig 已提交
772 773 774 775 776 777 778 779 780

	/*
	 * XXX(hch): Updating the ACL entries is not atomic vs the i_mode
	 * 	     update.  We could avoid this with linked transactions
	 * 	     and passing down the transaction pointer all the way
	 *	     to attr_set.  No previous user of the generic
	 * 	     Posix ACL code seems to care about this issue either.
	 */
	if ((mask & ATTR_MODE) && !(flags & XFS_ATTR_NOACL)) {
D
Dave Chinner 已提交
781
		error = posix_acl_chmod(inode, inode->i_mode);
C
Christoph Hellwig 已提交
782
		if (error)
E
Eric Sandeen 已提交
783
			return error;
C
Christoph Hellwig 已提交
784 785 786 787
	}

	return 0;

788
out_cancel:
789
	xfs_trans_cancel(tp);
790
out_dqrele:
C
Christoph Hellwig 已提交
791 792 793 794 795
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	return error;
}

796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
int
xfs_vn_setattr_nonsize(
	struct dentry		*dentry,
	struct iattr		*iattr)
{
	struct xfs_inode	*ip = XFS_I(d_inode(dentry));
	int error;

	trace_xfs_setattr(ip);

	error = xfs_vn_change_ok(dentry, iattr);
	if (error)
		return error;
	return xfs_setattr_nonsize(ip, iattr, 0);
}

C
Christoph Hellwig 已提交
812 813
/*
 * Truncate file.  Must have write permission and not be a directory.
814 815
 *
 * Caution: The caller of this function is responsible for calling
816
 * setattr_prepare() or otherwise verifying the change is fine.
C
Christoph Hellwig 已提交
817
 */
D
Darrick J. Wong 已提交
818
STATIC int
C
Christoph Hellwig 已提交
819 820
xfs_setattr_size(
	struct xfs_inode	*ip,
821
	struct iattr		*iattr)
C
Christoph Hellwig 已提交
822 823 824
{
	struct xfs_mount	*mp = ip->i_mount;
	struct inode		*inode = VFS_I(ip);
825
	xfs_off_t		oldsize, newsize;
C
Christoph Hellwig 已提交
826 827
	struct xfs_trans	*tp;
	int			error;
828
	uint			lock_flags = 0;
829
	bool			did_zeroing = false;
C
Christoph Hellwig 已提交
830

831
	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
832
	ASSERT(xfs_isilocked(ip, XFS_MMAPLOCK_EXCL));
D
Dave Chinner 已提交
833
	ASSERT(S_ISREG(inode->i_mode));
834 835
	ASSERT((iattr->ia_valid & (ATTR_UID|ATTR_GID|ATTR_ATIME|ATTR_ATIME_SET|
		ATTR_MTIME_SET|ATTR_KILL_PRIV|ATTR_TIMES_SET)) == 0);
C
Christoph Hellwig 已提交
836

837
	oldsize = inode->i_size;
838 839
	newsize = iattr->ia_size;

C
Christoph Hellwig 已提交
840 841 842
	/*
	 * Short circuit the truncate case for zero length files.
	 */
843
	if (newsize == 0 && oldsize == 0 && ip->i_d.di_nextents == 0) {
844
		if (!(iattr->ia_valid & (ATTR_CTIME|ATTR_MTIME)))
845
			return 0;
846 847 848 849 850 851

		/*
		 * Use the regular setattr path to update the timestamps.
		 */
		iattr->ia_valid &= ~ATTR_SIZE;
		return xfs_setattr_nonsize(ip, iattr, 0);
C
Christoph Hellwig 已提交
852 853 854 855 856
	}

	/*
	 * Make sure that the dquots are attached to the inode.
	 */
857
	error = xfs_qm_dqattach(ip, 0);
C
Christoph Hellwig 已提交
858
	if (error)
859
		return error;
C
Christoph Hellwig 已提交
860

861 862 863 864 865
	/*
	 * Wait for all direct I/O to complete.
	 */
	inode_dio_wait(inode);

C
Christoph Hellwig 已提交
866
	/*
867 868 869 870 871
	 * File data changes must be complete before we start the transaction to
	 * modify the inode.  This needs to be done before joining the inode to
	 * the transaction because the inode cannot be unlocked once it is a
	 * part of the transaction.
	 *
872 873 874
	 * Start with zeroing any data beyond EOF that we may expose on file
	 * extension, or zeroing out the rest of the block on a downward
	 * truncate.
C
Christoph Hellwig 已提交
875
	 */
876
	if (newsize > oldsize) {
877
		error = xfs_zero_eof(ip, newsize, oldsize, &did_zeroing);
878
	} else {
879 880
		error = iomap_truncate_page(inode, newsize, &did_zeroing,
				&xfs_iomap_ops);
C
Christoph Hellwig 已提交
881 882
	}

883 884 885
	if (error)
		return error;

C
Christoph Hellwig 已提交
886 887 888 889 890
	/*
	 * We are going to log the inode size change in this transaction so
	 * any previous writes that are beyond the on disk EOF and the new
	 * EOF that have not been written out need to be written here.  If we
	 * do not write the data out, we expose ourselves to the null files
891 892
	 * problem. Note that this includes any block zeroing we did above;
	 * otherwise those blocks may not be zeroed after a crash.
C
Christoph Hellwig 已提交
893
	 */
894 895
	if (did_zeroing ||
	    (newsize > ip->i_d.di_size && oldsize != ip->i_d.di_size)) {
D
Dave Chinner 已提交
896
		error = filemap_write_and_wait_range(VFS_I(ip)->i_mapping,
D
Dave Chinner 已提交
897
						      ip->i_d.di_size, newsize);
C
Christoph Hellwig 已提交
898
		if (error)
899
			return error;
C
Christoph Hellwig 已提交
900 901
	}

902
	/*
903 904 905 906 907 908
	 * We've already locked out new page faults, so now we can safely remove
	 * pages from the page cache knowing they won't get refaulted until we
	 * drop the XFS_MMAP_EXCL lock after the extent manipulations are
	 * complete. The truncate_setsize() call also cleans partial EOF page
	 * PTEs on extending truncates and hence ensures sub-page block size
	 * filesystems are correctly handled, too.
909
	 *
910 911 912
	 * We have to do all the page cache truncate work outside the
	 * transaction context as the "lock" order is page lock->log space
	 * reservation as defined by extent allocation in the writeback path.
913
	 * Hence a truncate can fail with ENOMEM from xfs_trans_alloc(), but
914 915 916 917
	 * having already truncated the in-memory version of the file (i.e. made
	 * user visible changes). There's not much we can do about this, except
	 * to hope that the caller sees ENOMEM and retries the truncate
	 * operation.
918 919
	 */
	truncate_setsize(inode, newsize);
C
Christoph Hellwig 已提交
920

921
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate, 0, 0, 0, &tp);
C
Christoph Hellwig 已提交
922
	if (error)
923
		return error;
C
Christoph Hellwig 已提交
924 925 926

	lock_flags |= XFS_ILOCK_EXCL;
	xfs_ilock(ip, XFS_ILOCK_EXCL);
927
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
928 929 930 931 932 933 934 935 936 937 938

	/*
	 * Only change the c/mtime if we are changing the size or we are
	 * explicitly asked to change it.  This handles the semantic difference
	 * between truncate() and ftruncate() as implemented in the VFS.
	 *
	 * The regular truncate() case without ATTR_CTIME and ATTR_MTIME is a
	 * special case where we need to update the times despite not having
	 * these flags set.  For all other operations the VFS set these flags
	 * explicitly if it wants a timestamp update.
	 */
939 940
	if (newsize != oldsize &&
	    !(iattr->ia_valid & (ATTR_CTIME | ATTR_MTIME))) {
C
Christoph Hellwig 已提交
941
		iattr->ia_ctime = iattr->ia_mtime =
942
			current_time(inode);
943
		iattr->ia_valid |= ATTR_CTIME | ATTR_MTIME;
C
Christoph Hellwig 已提交
944 945
	}

946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962
	/*
	 * The first thing we do is set the size to new_size permanently on
	 * disk.  This way we don't have to worry about anyone ever being able
	 * to look at the data being freed even in the face of a crash.
	 * What we're getting around here is the case where we free a block, it
	 * is allocated to another file, it is written to, and then we crash.
	 * If the new data gets written to the file but the log buffers
	 * containing the free and reallocation don't, then we'd end up with
	 * garbage in the blocks being freed.  As long as we make the new size
	 * permanent before actually freeing any blocks it doesn't matter if
	 * they get written to.
	 */
	ip->i_d.di_size = newsize;
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	if (newsize <= oldsize) {
		error = xfs_itruncate_extents(&tp, ip, XFS_DATA_FORK, newsize);
C
Christoph Hellwig 已提交
963
		if (error)
964
			goto out_trans_cancel;
C
Christoph Hellwig 已提交
965 966 967 968 969 970 971 972 973

		/*
		 * Truncated "down", so we're removing references to old data
		 * here - if we delay flushing for a long time, we expose
		 * ourselves unduly to the notorious NULL files problem.  So,
		 * we mark this inode and flush it when the file is closed,
		 * and do not wait the usual (long) time for writeout.
		 */
		xfs_iflags_set(ip, XFS_ITRUNCATED);
974 975 976

		/* A truncate down always removes post-EOF blocks. */
		xfs_inode_clear_eofblocks_tag(ip);
C
Christoph Hellwig 已提交
977 978
	}

979
	if (iattr->ia_valid & ATTR_MODE)
980
		xfs_setattr_mode(ip, iattr);
981
	if (iattr->ia_valid & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
C
Christoph Hellwig 已提交
982
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
983 984 985

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

986
	XFS_STATS_INC(mp, xs_ig_attrchg);
C
Christoph Hellwig 已提交
987 988 989 990

	if (mp->m_flags & XFS_MOUNT_WSYNC)
		xfs_trans_set_sync(tp);

991
	error = xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
992 993 994 995 996 997
out_unlock:
	if (lock_flags)
		xfs_iunlock(ip, lock_flags);
	return error;

out_trans_cancel:
998
	xfs_trans_cancel(tp);
C
Christoph Hellwig 已提交
999 1000 1001
	goto out_unlock;
}

1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
int
xfs_vn_setattr_size(
	struct dentry		*dentry,
	struct iattr		*iattr)
{
	struct xfs_inode	*ip = XFS_I(d_inode(dentry));
	int error;

	trace_xfs_setattr(ip);

	error = xfs_vn_change_ok(dentry, iattr);
	if (error)
		return error;
	return xfs_setattr_size(ip, iattr);
}

L
Linus Torvalds 已提交
1018
STATIC int
1019
xfs_vn_setattr(
1020 1021
	struct dentry		*dentry,
	struct iattr		*iattr)
L
Linus Torvalds 已提交
1022
{
1023 1024 1025
	int			error;

	if (iattr->ia_valid & ATTR_SIZE) {
1026 1027
		struct xfs_inode	*ip = XFS_I(d_inode(dentry));
		uint			iolock = XFS_IOLOCK_EXCL;
1028

1029 1030 1031
		error = xfs_break_layouts(d_inode(dentry), &iolock);
		if (error)
			return error;
1032

1033
		xfs_ilock(ip, XFS_MMAPLOCK_EXCL);
1034
		error = xfs_vn_setattr_size(dentry, iattr);
1035
		xfs_iunlock(ip, XFS_MMAPLOCK_EXCL);
1036
	} else {
1037
		error = xfs_vn_setattr_nonsize(dentry, iattr);
1038 1039
	}

D
Dave Chinner 已提交
1040
	return error;
L
Linus Torvalds 已提交
1041 1042
}

C
Christoph Hellwig 已提交
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
STATIC int
xfs_vn_update_time(
	struct inode		*inode,
	struct timespec		*now,
	int			flags)
{
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_trans	*tp;
	int			error;

	trace_xfs_update_time(ip);

1056 1057
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
	if (error)
D
Dave Chinner 已提交
1058
		return error;
C
Christoph Hellwig 已提交
1059 1060

	xfs_ilock(ip, XFS_ILOCK_EXCL);
1061
	if (flags & S_CTIME)
C
Christoph Hellwig 已提交
1062
		inode->i_ctime = *now;
1063
	if (flags & S_MTIME)
C
Christoph Hellwig 已提交
1064
		inode->i_mtime = *now;
1065
	if (flags & S_ATIME)
C
Christoph Hellwig 已提交
1066
		inode->i_atime = *now;
1067

C
Christoph Hellwig 已提交
1068 1069
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
	xfs_trans_log_inode(tp, ip, XFS_ILOG_TIMESTAMP);
1070
	return xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
1071 1072
}

E
Eric Sandeen 已提交
1073 1074 1075 1076 1077 1078 1079 1080 1081
STATIC int
xfs_vn_fiemap(
	struct inode		*inode,
	struct fiemap_extent_info *fieinfo,
	u64			start,
	u64			length)
{
	int			error;

1082
	xfs_ilock(XFS_I(inode), XFS_IOLOCK_SHARED);
1083 1084 1085 1086 1087 1088 1089 1090
	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
		fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR;
		error = iomap_fiemap(inode, fieinfo, start, length,
				&xfs_xattr_iomap_ops);
	} else {
		error = iomap_fiemap(inode, fieinfo, start, length,
				&xfs_iomap_ops);
	}
1091
	xfs_iunlock(XFS_I(inode), XFS_IOLOCK_SHARED);
E
Eric Sandeen 已提交
1092

1093
	return error;
E
Eric Sandeen 已提交
1094 1095
}

Z
Zhi Yong Wu 已提交
1096 1097 1098 1099 1100 1101
STATIC int
xfs_vn_tmpfile(
	struct inode	*dir,
	struct dentry	*dentry,
	umode_t		mode)
{
1102
	return xfs_generic_create(dir, dentry, mode, 0, true);
Z
Zhi Yong Wu 已提交
1103 1104
}

1105
static const struct inode_operations xfs_inode_operations = {
1106
	.get_acl		= xfs_get_acl,
1107
	.set_acl		= xfs_set_acl,
1108 1109 1110
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
E
Eric Sandeen 已提交
1111
	.fiemap			= xfs_vn_fiemap,
C
Christoph Hellwig 已提交
1112
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1113 1114
};

1115
static const struct inode_operations xfs_dir_inode_operations = {
1116 1117 1118 1119 1120 1121
	.create			= xfs_vn_create,
	.lookup			= xfs_vn_lookup,
	.link			= xfs_vn_link,
	.unlink			= xfs_vn_unlink,
	.symlink		= xfs_vn_symlink,
	.mkdir			= xfs_vn_mkdir,
1122 1123 1124 1125 1126 1127 1128
	/*
	 * Yes, XFS uses the same method for rmdir and unlink.
	 *
	 * There are some subtile differences deeper in the code,
	 * but we use S_ISDIR to check for those.
	 */
	.rmdir			= xfs_vn_unlink,
1129
	.mknod			= xfs_vn_mknod,
1130
	.rename			= xfs_vn_rename,
1131
	.get_acl		= xfs_get_acl,
1132
	.set_acl		= xfs_set_acl,
1133 1134 1135
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1136
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1137
	.tmpfile		= xfs_vn_tmpfile,
L
Linus Torvalds 已提交
1138 1139
};

1140
static const struct inode_operations xfs_dir_ci_inode_operations = {
1141 1142 1143 1144 1145 1146
	.create			= xfs_vn_create,
	.lookup			= xfs_vn_ci_lookup,
	.link			= xfs_vn_link,
	.unlink			= xfs_vn_unlink,
	.symlink		= xfs_vn_symlink,
	.mkdir			= xfs_vn_mkdir,
1147 1148 1149 1150 1151 1152 1153
	/*
	 * Yes, XFS uses the same method for rmdir and unlink.
	 *
	 * There are some subtile differences deeper in the code,
	 * but we use S_ISDIR to check for those.
	 */
	.rmdir			= xfs_vn_unlink,
1154
	.mknod			= xfs_vn_mknod,
1155
	.rename			= xfs_vn_rename,
1156
	.get_acl		= xfs_get_acl,
1157
	.set_acl		= xfs_set_acl,
1158 1159 1160
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1161
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1162
	.tmpfile		= xfs_vn_tmpfile,
1163 1164
};

1165
static const struct inode_operations xfs_symlink_inode_operations = {
1166
	.get_link		= xfs_vn_get_link,
1167 1168 1169
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1170
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1171
};
1172

1173 1174 1175 1176 1177 1178 1179 1180
static const struct inode_operations xfs_inline_symlink_inode_operations = {
	.get_link		= xfs_vn_get_link_inline,
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
	.update_time		= xfs_vn_update_time,
};

1181 1182 1183 1184 1185
STATIC void
xfs_diflags_to_iflags(
	struct inode		*inode,
	struct xfs_inode	*ip)
{
D
Dave Chinner 已提交
1186 1187 1188 1189 1190 1191
	uint16_t		flags = ip->i_d.di_flags;

	inode->i_flags &= ~(S_IMMUTABLE | S_APPEND | S_SYNC |
			    S_NOATIME | S_DAX);

	if (flags & XFS_DIFLAG_IMMUTABLE)
1192
		inode->i_flags |= S_IMMUTABLE;
D
Dave Chinner 已提交
1193
	if (flags & XFS_DIFLAG_APPEND)
1194
		inode->i_flags |= S_APPEND;
D
Dave Chinner 已提交
1195
	if (flags & XFS_DIFLAG_SYNC)
1196
		inode->i_flags |= S_SYNC;
D
Dave Chinner 已提交
1197
	if (flags & XFS_DIFLAG_NOATIME)
1198
		inode->i_flags |= S_NOATIME;
1199
	if (S_ISREG(inode->i_mode) &&
1200
	    ip->i_mount->m_sb.sb_blocksize == PAGE_SIZE &&
1201
	    !xfs_is_reflink_inode(ip) &&
1202 1203
	    (ip->i_mount->m_flags & XFS_MOUNT_DAX ||
	     ip->i_d.di_flags2 & XFS_DIFLAG2_DAX))
D
Dave Chinner 已提交
1204
		inode->i_flags |= S_DAX;
1205 1206 1207
}

/*
1208
 * Initialize the Linux inode.
1209
 *
1210 1211 1212 1213
 * When reading existing inodes from disk this is called directly from xfs_iget,
 * when creating a new inode it is called from xfs_ialloc after setting up the
 * inode. These callers have different criteria for clearing XFS_INEW, so leave
 * it up to the caller to deal with unlocking the inode appropriately.
1214 1215 1216 1217 1218
 */
void
xfs_setup_inode(
	struct xfs_inode	*ip)
{
1219
	struct inode		*inode = &ip->i_vnode;
1220
	gfp_t			gfp_mask;
1221 1222

	inode->i_ino = ip->i_ino;
C
Christoph Hellwig 已提交
1223
	inode->i_state = I_NEW;
1224 1225

	inode_sb_list_add(inode);
C
Christoph Hellwig 已提交
1226 1227
	/* make the inode look hashed for the writeback code */
	hlist_add_fake(&inode->i_hash);
1228

1229 1230
	inode->i_uid    = xfs_uid_to_kuid(ip->i_d.di_uid);
	inode->i_gid    = xfs_gid_to_kgid(ip->i_d.di_gid);
1231 1232 1233 1234

	i_size_write(inode, ip->i_d.di_size);
	xfs_diflags_to_iflags(inode, ip);

1235
	if (S_ISDIR(inode->i_mode)) {
1236
		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_dir_ilock_class);
1237
		ip->d_ops = ip->i_mount->m_dir_inode_ops;
1238 1239 1240
	} else {
		ip->d_ops = ip->i_mount->m_nondir_inode_ops;
		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_nondir_ilock_class);
1241 1242
	}

1243 1244 1245 1246 1247 1248 1249 1250
	/*
	 * Ensure all page cache allocations are done from GFP_NOFS context to
	 * prevent direct reclaim recursion back into the filesystem and blowing
	 * stacks or deadlocking.
	 */
	gfp_mask = mapping_gfp_mask(inode->i_mapping);
	mapping_set_gfp_mask(inode->i_mapping, (gfp_mask & ~(__GFP_FS)));

1251 1252 1253 1254 1255 1256
	/*
	 * If there is no attribute fork no ACL can exist on this inode,
	 * and it can't have any file capabilities attached to it either.
	 */
	if (!XFS_IFORK_Q(ip)) {
		inode_has_no_xattr(inode);
1257
		cache_no_acl(inode);
1258
	}
1259
}
1260 1261 1262 1263 1264 1265 1266

void
xfs_setup_iops(
	struct xfs_inode	*ip)
{
	struct inode		*inode = &ip->i_vnode;

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &xfs_inode_operations;
		inode->i_fop = &xfs_file_operations;
		inode->i_mapping->a_ops = &xfs_address_space_operations;
		break;
	case S_IFDIR:
		if (xfs_sb_version_hasasciici(&XFS_M(inode->i_sb)->m_sb))
			inode->i_op = &xfs_dir_ci_inode_operations;
		else
			inode->i_op = &xfs_dir_inode_operations;
		inode->i_fop = &xfs_dir_file_operations;
		break;
	case S_IFLNK:
1281 1282 1283 1284
		if (ip->i_df.if_flags & XFS_IFINLINE)
			inode->i_op = &xfs_inline_symlink_inode_operations;
		else
			inode->i_op = &xfs_symlink_inode_operations;
1285 1286 1287 1288 1289 1290 1291
		break;
	default:
		inode->i_op = &xfs_inode_operations;
		init_special_inode(inode, inode->i_mode, inode->i_rdev);
		break;
	}
}