xfs_iops.c 32.9 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>
49
#include <linux/iversion.h>
L
Linus Torvalds 已提交
50

51 52 53 54 55 56 57 58 59 60 61 62
/*
 * 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;

63 64 65 66 67
static int
xfs_initxattrs(
	struct inode		*inode,
	const struct xattr	*xattr_array,
	void			*fs_info)
68
{
69 70 71
	const struct xattr	*xattr;
	struct xfs_inode	*ip = XFS_I(inode);
	int			error = 0;
72 73

	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
D
Dave Chinner 已提交
74
		error = xfs_attr_set(ip, xattr->name, xattr->value,
75
				      xattr->value_len, ATTR_SECURE);
76 77 78 79 80 81
		if (error < 0)
			break;
	}
	return error;
}

82 83 84 85 86 87
/*
 * 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).
 */
88

89
STATIC int
90
xfs_init_security(
91
	struct inode	*inode,
92 93
	struct inode	*dir,
	const struct qstr *qstr)
94
{
D
Dave Chinner 已提交
95
	return security_inode_init_security(inode, dir, qstr,
96
					     &xfs_initxattrs, NULL);
97 98
}

99 100
static void
xfs_dentry_to_name(
101 102 103 104 105 106 107 108 109 110
	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(
111
	struct xfs_name	*namep,
112 113
	struct dentry	*dentry,
	int		mode)
114 115 116
{
	namep->name = dentry->d_name.name;
	namep->len = dentry->d_name.len;
117
	namep->type = xfs_mode_to_ftype(mode);
118 119 120 121 122

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

	return 0;
123 124
}

125
STATIC void
126
xfs_cleanup_inode(
127
	struct inode	*dir,
128
	struct inode	*inode,
129
	struct dentry	*dentry)
130
{
131
	struct xfs_name	teardown;
132 133

	/* Oh, the horror.
134
	 * If we can't add the ACL or we fail in
135
	 * xfs_init_security we must back out.
136 137
	 * ENOSPC can hit here, among other things.
	 */
138
	xfs_dentry_to_name(&teardown, dentry);
139

140
	xfs_remove(XFS_I(dir), &teardown, XFS_I(inode));
141 142
}

L
Linus Torvalds 已提交
143
STATIC int
144
xfs_generic_create(
L
Linus Torvalds 已提交
145 146
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
147
	umode_t		mode,
148 149
	dev_t		rdev,
	bool		tmpfile)	/* unnamed file */
L
Linus Torvalds 已提交
150
{
C
Christoph Hellwig 已提交
151
	struct inode	*inode;
152
	struct xfs_inode *ip = NULL;
153
	struct posix_acl *default_acl, *acl;
154
	struct xfs_name	name;
L
Linus Torvalds 已提交
155 156 157 158 159 160
	int		error;

	/*
	 * Irix uses Missed'em'V split, but doesn't want to see
	 * the upper 5 bits of (14bit) major.
	 */
161 162 163 164 165 166
	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
167

168 169 170
	error = posix_acl_create(dir, &mode, &default_acl, &acl);
	if (error)
		return error;
L
Linus Torvalds 已提交
171

172 173 174 175 176
	/* Verify mode is valid also for tmpfile case */
	error = xfs_dentry_mode_to_name(&name, dentry, mode);
	if (unlikely(error))
		goto out_free_acl;

177 178 179
	if (!tmpfile) {
		error = xfs_create(XFS_I(dir), &name, mode, rdev, &ip);
	} else {
180
		error = xfs_create_tmpfile(XFS_I(dir), mode, &ip);
181
	}
C
Christoph Hellwig 已提交
182 183
	if (unlikely(error))
		goto out_free_acl;
184

185
	inode = VFS_I(ip);
186

187
	error = xfs_init_security(inode, dir, &dentry->d_name);
C
Christoph Hellwig 已提交
188 189 190
	if (unlikely(error))
		goto out_cleanup_inode;

191
#ifdef CONFIG_XFS_POSIX_ACL
C
Christoph Hellwig 已提交
192
	if (default_acl) {
193
		error = __xfs_set_acl(inode, default_acl, ACL_TYPE_DEFAULT);
194
		if (error)
C
Christoph Hellwig 已提交
195
			goto out_cleanup_inode;
L
Linus Torvalds 已提交
196
	}
197
	if (acl) {
198
		error = __xfs_set_acl(inode, acl, ACL_TYPE_ACCESS);
199 200 201 202
		if (error)
			goto out_cleanup_inode;
	}
#endif
L
Linus Torvalds 已提交
203

204 205
	xfs_setup_iops(ip);

206 207 208 209 210
	if (tmpfile)
		d_tmpfile(dentry, inode);
	else
		d_instantiate(dentry, inode);

211 212
	xfs_finish_inode_setup(ip);

213 214 215 216 217
 out_free_acl:
	if (default_acl)
		posix_acl_release(default_acl);
	if (acl)
		posix_acl_release(acl);
D
Dave Chinner 已提交
218
	return error;
C
Christoph Hellwig 已提交
219 220

 out_cleanup_inode:
221
	xfs_finish_inode_setup(ip);
222 223 224
	if (!tmpfile)
		xfs_cleanup_inode(dir, inode, dentry);
	iput(inode);
225
	goto out_free_acl;
L
Linus Torvalds 已提交
226 227
}

228 229 230 231 232 233 234 235 236 237
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
238
STATIC int
239
xfs_vn_create(
L
Linus Torvalds 已提交
240 241
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
242
	umode_t		mode,
A
Al Viro 已提交
243
	bool		flags)
L
Linus Torvalds 已提交
244
{
245
	return xfs_vn_mknod(dir, dentry, mode, 0);
L
Linus Torvalds 已提交
246 247 248
}

STATIC int
249
xfs_vn_mkdir(
L
Linus Torvalds 已提交
250 251
	struct inode	*dir,
	struct dentry	*dentry,
252
	umode_t		mode)
L
Linus Torvalds 已提交
253
{
254
	return xfs_vn_mknod(dir, dentry, mode|S_IFDIR, 0);
L
Linus Torvalds 已提交
255 256 257
}

STATIC struct dentry *
258
xfs_vn_lookup(
L
Linus Torvalds 已提交
259 260
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
261
	unsigned int flags)
L
Linus Torvalds 已提交
262
{
263
	struct xfs_inode *cip;
264
	struct xfs_name	name;
L
Linus Torvalds 已提交
265 266 267 268 269
	int		error;

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

270
	xfs_dentry_to_name(&name, dentry);
271
	error = xfs_lookup(XFS_I(dir), &name, &cip, NULL);
272
	if (unlikely(error)) {
D
Dave Chinner 已提交
273 274
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
L
Linus Torvalds 已提交
275 276 277 278
		d_add(dentry, NULL);
		return NULL;
	}

279
	return d_splice_alias(VFS_I(cip), dentry);
L
Linus Torvalds 已提交
280 281
}

282 283 284 285
STATIC struct dentry *
xfs_vn_ci_lookup(
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
286
	unsigned int flags)
287 288 289 290 291 292 293 294 295 296
{
	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);

297
	xfs_dentry_to_name(&xname, dentry);
298 299
	error = xfs_lookup(XFS_I(dir), &xname, &ip, &ci_name);
	if (unlikely(error)) {
D
Dave Chinner 已提交
300 301
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
302 303 304 305 306
		/*
		 * call d_add(dentry, NULL) here when d_drop_negative_children
		 * is called in xfs_vn_mknod (ie. allow negative dentries
		 * with CI filesystems).
		 */
307 308 309 310 311
		return NULL;
	}

	/* if exact match, just splice and exit */
	if (!ci_name.name)
312
		return d_splice_alias(VFS_I(ip), dentry);
313 314 315 316

	/* else case-insensitive match... */
	dname.name = ci_name.name;
	dname.len = ci_name.len;
317
	dentry = d_add_ci(dentry, VFS_I(ip), &dname);
318 319 320 321
	kmem_free(ci_name.name);
	return dentry;
}

L
Linus Torvalds 已提交
322
STATIC int
323
xfs_vn_link(
L
Linus Torvalds 已提交
324 325 326 327
	struct dentry	*old_dentry,
	struct inode	*dir,
	struct dentry	*dentry)
{
328
	struct inode	*inode = d_inode(old_dentry);
329
	struct xfs_name	name;
L
Linus Torvalds 已提交
330 331
	int		error;

332 333 334
	error = xfs_dentry_mode_to_name(&name, dentry, inode->i_mode);
	if (unlikely(error))
		return error;
L
Linus Torvalds 已提交
335

336
	error = xfs_link(XFS_I(dir), XFS_I(inode), &name);
337
	if (unlikely(error))
D
Dave Chinner 已提交
338
		return error;
339

A
Al Viro 已提交
340
	ihold(inode);
341 342
	d_instantiate(dentry, inode);
	return 0;
L
Linus Torvalds 已提交
343 344 345
}

STATIC int
346
xfs_vn_unlink(
L
Linus Torvalds 已提交
347 348 349
	struct inode	*dir,
	struct dentry	*dentry)
{
350
	struct xfs_name	name;
L
Linus Torvalds 已提交
351 352
	int		error;

353
	xfs_dentry_to_name(&name, dentry);
L
Linus Torvalds 已提交
354

355
	error = xfs_remove(XFS_I(dir), &name, XFS_I(d_inode(dentry)));
356 357 358 359 360 361 362 363 364 365 366
	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
367 368 369
}

STATIC int
370
xfs_vn_symlink(
L
Linus Torvalds 已提交
371 372 373 374
	struct inode	*dir,
	struct dentry	*dentry,
	const char	*symname)
{
375 376
	struct inode	*inode;
	struct xfs_inode *cip = NULL;
377
	struct xfs_name	name;
L
Linus Torvalds 已提交
378
	int		error;
A
Al Viro 已提交
379
	umode_t		mode;
L
Linus Torvalds 已提交
380

381
	mode = S_IFLNK |
A
Al Viro 已提交
382
		(irix_symlink_mode ? 0777 & ~current_umask() : S_IRWXUGO);
383 384 385
	error = xfs_dentry_mode_to_name(&name, dentry, mode);
	if (unlikely(error))
		goto out;
L
Linus Torvalds 已提交
386

C
Christoph Hellwig 已提交
387
	error = xfs_symlink(XFS_I(dir), &name, symname, mode, &cip);
388 389 390
	if (unlikely(error))
		goto out;

391
	inode = VFS_I(cip);
392

393
	error = xfs_init_security(inode, dir, &dentry->d_name);
394 395 396
	if (unlikely(error))
		goto out_cleanup_inode;

397 398
	xfs_setup_iops(cip);

399
	d_instantiate(dentry, inode);
400
	xfs_finish_inode_setup(cip);
401 402 403
	return 0;

 out_cleanup_inode:
404
	xfs_finish_inode_setup(cip);
405
	xfs_cleanup_inode(dir, inode, dentry);
406
	iput(inode);
407
 out:
D
Dave Chinner 已提交
408
	return error;
L
Linus Torvalds 已提交
409 410 411
}

STATIC int
412
xfs_vn_rename(
L
Linus Torvalds 已提交
413 414 415
	struct inode	*odir,
	struct dentry	*odentry,
	struct inode	*ndir,
416 417
	struct dentry	*ndentry,
	unsigned int	flags)
L
Linus Torvalds 已提交
418
{
419
	struct inode	*new_inode = d_inode(ndentry);
420
	int		omode = 0;
421
	int		error;
422 423
	struct xfs_name	oname;
	struct xfs_name	nname;
L
Linus Torvalds 已提交
424

D
Dave Chinner 已提交
425
	if (flags & ~(RENAME_NOREPLACE | RENAME_EXCHANGE | RENAME_WHITEOUT))
426 427
		return -EINVAL;

428 429
	/* if we are exchanging files, we need to set i_mode of both files */
	if (flags & RENAME_EXCHANGE)
430
		omode = d_inode(ndentry)->i_mode;
431

432 433 434 435 436 437 438 439
	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;
440

441
	return xfs_rename(XFS_I(odir), &oname, XFS_I(d_inode(odentry)),
442
			  XFS_I(ndir), &nname,
443
			  new_inode ? XFS_I(new_inode) : NULL, flags);
L
Linus Torvalds 已提交
444 445 446 447 448 449 450
}

/*
 * 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...
 */
451
STATIC const char *
452
xfs_vn_get_link(
L
Linus Torvalds 已提交
453
	struct dentry		*dentry,
454
	struct inode		*inode,
455
	struct delayed_call	*done)
L
Linus Torvalds 已提交
456 457
{
	char			*link;
458
	int			error = -ENOMEM;
L
Linus Torvalds 已提交
459

460 461 462
	if (!dentry)
		return ERR_PTR(-ECHILD);

463
	link = kmalloc(XFS_SYMLINK_MAXLEN+1, GFP_KERNEL);
464 465
	if (!link)
		goto out_err;
L
Linus Torvalds 已提交
466

467
	error = xfs_readlink(XFS_I(d_inode(dentry)), link);
468 469
	if (unlikely(error))
		goto out_kfree;
L
Linus Torvalds 已提交
470

471 472
	set_delayed_call(done, kfree_link, link);
	return link;
473 474 475 476

 out_kfree:
	kfree(link);
 out_err:
477
	return ERR_PTR(error);
L
Linus Torvalds 已提交
478 479
}

480 481 482 483 484 485 486 487 488 489
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
490
STATIC int
491
xfs_vn_getattr(
492 493 494 495
	const struct path	*path,
	struct kstat		*stat,
	u32			request_mask,
	unsigned int		query_flags)
L
Linus Torvalds 已提交
496
{
497
	struct inode		*inode = d_inode(path->dentry);
498 499 500
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;

C
Christoph Hellwig 已提交
501
	trace_xfs_getattr(ip);
502 503

	if (XFS_FORCED_SHUTDOWN(mp))
E
Eric Sandeen 已提交
504
		return -EIO;
505 506 507

	stat->size = XFS_ISIZE(ip);
	stat->dev = inode->i_sb->s_dev;
D
Dave Chinner 已提交
508
	stat->mode = inode->i_mode;
509
	stat->nlink = inode->i_nlink;
510 511
	stat->uid = inode->i_uid;
	stat->gid = inode->i_gid;
512 513
	stat->ino = ip->i_ino;
	stat->atime = inode->i_atime;
514 515
	stat->mtime = inode->i_mtime;
	stat->ctime = inode->i_ctime;
516 517 518
	stat->blocks =
		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);

519 520 521 522 523 524 525 526 527 528 529 530 531 532
	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;
533 534 535 536 537

	switch (inode->i_mode & S_IFMT) {
	case S_IFBLK:
	case S_IFCHR:
		stat->blksize = BLKDEV_IOSIZE;
C
Christoph Hellwig 已提交
538
		stat->rdev = inode->i_rdev;
539 540
		break;
	default:
541
		if (XFS_IS_REALTIME_INODE(ip)) {
542 543 544 545 546 547 548 549 550 551 552
			/*
			 * 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;
553
	}
554 555

	return 0;
L
Linus Torvalds 已提交
556 557
}

558 559 560 561 562
static void
xfs_setattr_mode(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
563 564
	struct inode		*inode = VFS_I(ip);
	umode_t			mode = iattr->ia_mode;
565 566 567 568 569 570 571

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

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

572
void
C
Christoph Hellwig 已提交
573 574 575 576 577 578 579 580
xfs_setattr_time(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
	struct inode		*inode = VFS_I(ip);

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

581
	if (iattr->ia_valid & ATTR_ATIME)
C
Christoph Hellwig 已提交
582
		inode->i_atime = iattr->ia_atime;
583
	if (iattr->ia_valid & ATTR_CTIME)
C
Christoph Hellwig 已提交
584
		inode->i_ctime = iattr->ia_ctime;
585
	if (iattr->ia_valid & ATTR_MTIME)
C
Christoph Hellwig 已提交
586 587 588
		inode->i_mtime = iattr->ia_mtime;
}

589 590 591 592 593
static int
xfs_vn_change_ok(
	struct dentry	*dentry,
	struct iattr	*iattr)
{
594
	struct xfs_mount	*mp = XFS_I(d_inode(dentry))->i_mount;
595 596 597 598 599 600 601

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

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

602
	return setattr_prepare(dentry, iattr);
603 604 605 606 607 608
}

/*
 * Set non-size attributes of an inode.
 *
 * Caution: The caller of this function is responsible for calling
609
 * setattr_prepare() or otherwise verifying the change is fine.
610
 */
C
Christoph Hellwig 已提交
611 612 613 614 615 616 617 618 619 620 621
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;
622 623
	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
624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643
	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 {
644
			uid = inode->i_uid;
C
Christoph Hellwig 已提交
645 646 647 648 649
		}
		if ((mask & ATTR_GID) && XFS_IS_GQUOTA_ON(mp)) {
			gid = iattr->ia_gid;
			qflags |= XFS_QMOPT_GQUOTA;
		}  else {
650
			gid = inode->i_gid;
C
Christoph Hellwig 已提交
651 652 653 654 655 656 657 658 659
		}

		/*
		 * 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);
660 661 662 663
		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
664 665 666 667
		if (error)
			return error;
	}

668
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ichange, 0, 0, 0, &tp);
C
Christoph Hellwig 已提交
669
	if (error)
670
		goto out_dqrele;
C
Christoph Hellwig 已提交
671 672

	xfs_ilock(ip, XFS_ILOCK_EXCL);
673
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
674 675 676 677 678 679 680 681 682 683 684

	/*
	 * 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.
		 */
685 686
		iuid = inode->i_uid;
		igid = inode->i_gid;
C
Christoph Hellwig 已提交
687 688 689 690 691 692 693 694
		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) &&
695 696
		    ((XFS_IS_UQUOTA_ON(mp) && !uid_eq(iuid, uid)) ||
		     (XFS_IS_GQUOTA_ON(mp) && !gid_eq(igid, gid)))) {
C
Christoph Hellwig 已提交
697 698
			ASSERT(tp);
			error = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
699
						NULL, capable(CAP_FOWNER) ?
C
Christoph Hellwig 已提交
700 701
						XFS_QMOPT_FORCE_RES : 0);
			if (error)	/* out of quota */
702
				goto out_cancel;
C
Christoph Hellwig 已提交
703 704 705 706 707 708 709 710 711 712 713 714 715
		}
	}

	/*
	 * 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
716
		if ((inode->i_mode & (S_ISUID|S_ISGID)) &&
C
Christoph Hellwig 已提交
717
		    !capable(CAP_FSETID))
D
Dave Chinner 已提交
718
			inode->i_mode &= ~(S_ISUID|S_ISGID);
C
Christoph Hellwig 已提交
719 720 721 722 723

		/*
		 * Change the ownerships and register quota modifications
		 * in the transaction.
		 */
724
		if (!uid_eq(iuid, uid)) {
C
Christoph Hellwig 已提交
725 726 727 728 729 730
			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);
			}
731
			ip->i_d.di_uid = xfs_kuid_to_uid(uid);
C
Christoph Hellwig 已提交
732 733
			inode->i_uid = uid;
		}
734
		if (!gid_eq(igid, gid)) {
C
Christoph Hellwig 已提交
735
			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_GQUOTA_ON(mp)) {
736 737
				ASSERT(xfs_sb_version_has_pquotino(&mp->m_sb) ||
				       !XFS_IS_PQUOTA_ON(mp));
C
Christoph Hellwig 已提交
738 739 740 741 742
				ASSERT(mask & ATTR_GID);
				ASSERT(gdqp);
				olddquot2 = xfs_qm_vop_chown(tp, ip,
							&ip->i_gdquot, gdqp);
			}
743
			ip->i_d.di_gid = xfs_kgid_to_gid(gid);
C
Christoph Hellwig 已提交
744 745 746 747
			inode->i_gid = gid;
		}
	}

748
	if (mask & ATTR_MODE)
749
		xfs_setattr_mode(ip, iattr);
C
Christoph Hellwig 已提交
750 751
	if (mask & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
752 753 754

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

755
	XFS_STATS_INC(mp, xs_ig_attrchg);
C
Christoph Hellwig 已提交
756 757 758

	if (mp->m_flags & XFS_MOUNT_WSYNC)
		xfs_trans_set_sync(tp);
759
	error = xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
760 761 762 763 764 765 766 767 768 769 770 771

	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
772
		return error;
C
Christoph Hellwig 已提交
773 774 775 776 777 778 779 780 781

	/*
	 * 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
782
		error = posix_acl_chmod(inode, inode->i_mode);
C
Christoph Hellwig 已提交
783
		if (error)
E
Eric Sandeen 已提交
784
			return error;
C
Christoph Hellwig 已提交
785 786 787 788
	}

	return 0;

789
out_cancel:
790
	xfs_trans_cancel(tp);
791
out_dqrele:
C
Christoph Hellwig 已提交
792 793 794 795 796
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	return error;
}

797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
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
813 814
/*
 * Truncate file.  Must have write permission and not be a directory.
815 816
 *
 * Caution: The caller of this function is responsible for calling
817
 * setattr_prepare() or otherwise verifying the change is fine.
C
Christoph Hellwig 已提交
818
 */
D
Darrick J. Wong 已提交
819
STATIC int
C
Christoph Hellwig 已提交
820 821
xfs_setattr_size(
	struct xfs_inode	*ip,
822
	struct iattr		*iattr)
C
Christoph Hellwig 已提交
823 824 825
{
	struct xfs_mount	*mp = ip->i_mount;
	struct inode		*inode = VFS_I(ip);
826
	xfs_off_t		oldsize, newsize;
C
Christoph Hellwig 已提交
827 828
	struct xfs_trans	*tp;
	int			error;
829
	uint			lock_flags = 0;
830
	bool			did_zeroing = false;
C
Christoph Hellwig 已提交
831

832
	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
833
	ASSERT(xfs_isilocked(ip, XFS_MMAPLOCK_EXCL));
D
Dave Chinner 已提交
834
	ASSERT(S_ISREG(inode->i_mode));
835 836
	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
837

838
	oldsize = inode->i_size;
839 840
	newsize = iattr->ia_size;

C
Christoph Hellwig 已提交
841 842 843
	/*
	 * Short circuit the truncate case for zero length files.
	 */
844
	if (newsize == 0 && oldsize == 0 && ip->i_d.di_nextents == 0) {
845
		if (!(iattr->ia_valid & (ATTR_CTIME|ATTR_MTIME)))
846
			return 0;
847 848 849 850 851 852

		/*
		 * 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
853 854 855 856 857
	}

	/*
	 * Make sure that the dquots are attached to the inode.
	 */
858
	error = xfs_qm_dqattach(ip);
C
Christoph Hellwig 已提交
859
	if (error)
860
		return error;
C
Christoph Hellwig 已提交
861

862 863 864 865 866
	/*
	 * Wait for all direct I/O to complete.
	 */
	inode_dio_wait(inode);

C
Christoph Hellwig 已提交
867
	/*
868 869 870 871 872
	 * 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.
	 *
873 874 875
	 * 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
876
	 */
877
	if (newsize > oldsize) {
C
Christoph Hellwig 已提交
878 879 880
		trace_xfs_zero_eof(ip, oldsize, newsize - oldsize);
		error = iomap_zero_range(inode, oldsize, newsize - oldsize,
				&did_zeroing, &xfs_iomap_ops);
881
	} else {
882 883
		error = iomap_truncate_page(inode, newsize, &did_zeroing,
				&xfs_iomap_ops);
C
Christoph Hellwig 已提交
884 885
	}

886 887 888
	if (error)
		return error;

889
	/*
890 891 892 893 894 895
	 * 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.
896
	 *
897 898 899
	 * 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.
900
	 * Hence a truncate can fail with ENOMEM from xfs_trans_alloc(), but
901 902 903 904
	 * 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.
905 906 907 908
	 *
	 * And we update in-core i_size and truncate page cache beyond newsize
	 * before writeback the [di_size, newsize] range, so we're guaranteed
	 * not to write stale data past the new EOF on truncate down.
909 910
	 */
	truncate_setsize(inode, newsize);
C
Christoph Hellwig 已提交
911

912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927
	/*
	 * 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
	 * problem. Note that this includes any block zeroing we did above;
	 * otherwise those blocks may not be zeroed after a crash.
	 */
	if (did_zeroing ||
	    (newsize > ip->i_d.di_size && oldsize != ip->i_d.di_size)) {
		error = filemap_write_and_wait_range(VFS_I(ip)->i_mapping,
						ip->i_d.di_size, newsize - 1);
		if (error)
			return error;
	}

928
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate, 0, 0, 0, &tp);
C
Christoph Hellwig 已提交
929
	if (error)
930
		return error;
C
Christoph Hellwig 已提交
931 932 933

	lock_flags |= XFS_ILOCK_EXCL;
	xfs_ilock(ip, XFS_ILOCK_EXCL);
934
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
935 936 937 938 939 940 941 942 943 944 945

	/*
	 * 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.
	 */
946 947
	if (newsize != oldsize &&
	    !(iattr->ia_valid & (ATTR_CTIME | ATTR_MTIME))) {
C
Christoph Hellwig 已提交
948
		iattr->ia_ctime = iattr->ia_mtime =
949
			current_time(inode);
950
		iattr->ia_valid |= ATTR_CTIME | ATTR_MTIME;
C
Christoph Hellwig 已提交
951 952
	}

953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969
	/*
	 * 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
970
		if (error)
971
			goto out_trans_cancel;
C
Christoph Hellwig 已提交
972 973 974 975 976 977 978 979 980

		/*
		 * 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);
981 982 983

		/* A truncate down always removes post-EOF blocks. */
		xfs_inode_clear_eofblocks_tag(ip);
C
Christoph Hellwig 已提交
984 985
	}

986
	if (iattr->ia_valid & ATTR_MODE)
987
		xfs_setattr_mode(ip, iattr);
988
	if (iattr->ia_valid & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
C
Christoph Hellwig 已提交
989
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
990 991 992

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

993
	XFS_STATS_INC(mp, xs_ig_attrchg);
C
Christoph Hellwig 已提交
994 995 996 997

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

998
	error = xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
999 1000 1001 1002 1003 1004
out_unlock:
	if (lock_flags)
		xfs_iunlock(ip, lock_flags);
	return error;

out_trans_cancel:
1005
	xfs_trans_cancel(tp);
C
Christoph Hellwig 已提交
1006 1007 1008
	goto out_unlock;
}

1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
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
1025
STATIC int
1026
xfs_vn_setattr(
1027 1028
	struct dentry		*dentry,
	struct iattr		*iattr)
L
Linus Torvalds 已提交
1029
{
1030 1031 1032
	int			error;

	if (iattr->ia_valid & ATTR_SIZE) {
1033 1034
		struct xfs_inode	*ip = XFS_I(d_inode(dentry));
		uint			iolock = XFS_IOLOCK_EXCL;
1035

1036 1037 1038
		error = xfs_break_layouts(d_inode(dentry), &iolock);
		if (error)
			return error;
1039

1040
		xfs_ilock(ip, XFS_MMAPLOCK_EXCL);
1041
		error = xfs_vn_setattr_size(dentry, iattr);
1042
		xfs_iunlock(ip, XFS_MMAPLOCK_EXCL);
1043
	} else {
1044
		error = xfs_vn_setattr_nonsize(dentry, iattr);
1045 1046
	}

D
Dave Chinner 已提交
1047
	return error;
L
Linus Torvalds 已提交
1048 1049
}

C
Christoph Hellwig 已提交
1050 1051 1052 1053 1054 1055 1056 1057
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;
1058
	int			log_flags = XFS_ILOG_TIMESTAMP;
C
Christoph Hellwig 已提交
1059 1060 1061 1062 1063
	struct xfs_trans	*tp;
	int			error;

	trace_xfs_update_time(ip);

1064 1065 1066 1067 1068 1069 1070 1071 1072
	if (inode->i_sb->s_flags & SB_LAZYTIME) {
		if (!((flags & S_VERSION) &&
		      inode_maybe_inc_iversion(inode, false)))
			return generic_update_time(inode, now, flags);

		/* Capture the iversion update that just occurred */
		log_flags |= XFS_ILOG_CORE;
	}

1073 1074
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
	if (error)
D
Dave Chinner 已提交
1075
		return error;
C
Christoph Hellwig 已提交
1076 1077

	xfs_ilock(ip, XFS_ILOCK_EXCL);
1078
	if (flags & S_CTIME)
C
Christoph Hellwig 已提交
1079
		inode->i_ctime = *now;
1080
	if (flags & S_MTIME)
C
Christoph Hellwig 已提交
1081
		inode->i_mtime = *now;
1082
	if (flags & S_ATIME)
C
Christoph Hellwig 已提交
1083
		inode->i_atime = *now;
1084

C
Christoph Hellwig 已提交
1085
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
1086
	xfs_trans_log_inode(tp, ip, log_flags);
1087
	return xfs_trans_commit(tp);
C
Christoph Hellwig 已提交
1088 1089
}

E
Eric Sandeen 已提交
1090 1091 1092 1093 1094 1095 1096 1097 1098
STATIC int
xfs_vn_fiemap(
	struct inode		*inode,
	struct fiemap_extent_info *fieinfo,
	u64			start,
	u64			length)
{
	int			error;

1099
	xfs_ilock(XFS_I(inode), XFS_IOLOCK_SHARED);
1100 1101 1102 1103 1104 1105 1106 1107
	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);
	}
1108
	xfs_iunlock(XFS_I(inode), XFS_IOLOCK_SHARED);
E
Eric Sandeen 已提交
1109

1110
	return error;
E
Eric Sandeen 已提交
1111 1112
}

Z
Zhi Yong Wu 已提交
1113 1114 1115 1116 1117 1118
STATIC int
xfs_vn_tmpfile(
	struct inode	*dir,
	struct dentry	*dentry,
	umode_t		mode)
{
1119
	return xfs_generic_create(dir, dentry, mode, 0, true);
Z
Zhi Yong Wu 已提交
1120 1121
}

1122
static const struct inode_operations xfs_inode_operations = {
1123
	.get_acl		= xfs_get_acl,
1124
	.set_acl		= xfs_set_acl,
1125 1126 1127
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
E
Eric Sandeen 已提交
1128
	.fiemap			= xfs_vn_fiemap,
C
Christoph Hellwig 已提交
1129
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1130 1131
};

1132
static const struct inode_operations xfs_dir_inode_operations = {
1133 1134 1135 1136 1137 1138
	.create			= xfs_vn_create,
	.lookup			= xfs_vn_lookup,
	.link			= xfs_vn_link,
	.unlink			= xfs_vn_unlink,
	.symlink		= xfs_vn_symlink,
	.mkdir			= xfs_vn_mkdir,
1139 1140 1141 1142 1143 1144 1145
	/*
	 * 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,
1146
	.mknod			= xfs_vn_mknod,
1147
	.rename			= xfs_vn_rename,
1148
	.get_acl		= xfs_get_acl,
1149
	.set_acl		= xfs_set_acl,
1150 1151 1152
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1153
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1154
	.tmpfile		= xfs_vn_tmpfile,
L
Linus Torvalds 已提交
1155 1156
};

1157
static const struct inode_operations xfs_dir_ci_inode_operations = {
1158 1159 1160 1161 1162 1163
	.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,
1164 1165 1166 1167 1168 1169 1170
	/*
	 * 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,
1171
	.mknod			= xfs_vn_mknod,
1172
	.rename			= xfs_vn_rename,
1173
	.get_acl		= xfs_get_acl,
1174
	.set_acl		= xfs_set_acl,
1175 1176 1177
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1178
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1179
	.tmpfile		= xfs_vn_tmpfile,
1180 1181
};

1182
static const struct inode_operations xfs_symlink_inode_operations = {
1183
	.get_link		= xfs_vn_get_link,
1184 1185 1186
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1187
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1188
};
1189

1190 1191 1192 1193 1194 1195 1196 1197
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,
};

1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
/* Figure out if this file actually supports DAX. */
static bool
xfs_inode_supports_dax(
	struct xfs_inode	*ip)
{
	struct xfs_mount	*mp = ip->i_mount;

	/* Only supported on non-reflinked files. */
	if (!S_ISREG(VFS_I(ip)->i_mode) || xfs_is_reflink_inode(ip))
		return false;

	/* DAX mount option or DAX iflag must be set. */
	if (!(mp->m_flags & XFS_MOUNT_DAX) &&
	    !(ip->i_d.di_flags2 & XFS_DIFLAG2_DAX))
		return false;

	/* Block size must match page size */
	if (mp->m_sb.sb_blocksize != PAGE_SIZE)
		return false;

	/* Device has to support DAX too. */
	return xfs_find_daxdev_for_inode(VFS_I(ip)) != NULL;
}

1222 1223 1224 1225 1226
STATIC void
xfs_diflags_to_iflags(
	struct inode		*inode,
	struct xfs_inode	*ip)
{
D
Dave Chinner 已提交
1227 1228 1229 1230 1231 1232
	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)
1233
		inode->i_flags |= S_IMMUTABLE;
D
Dave Chinner 已提交
1234
	if (flags & XFS_DIFLAG_APPEND)
1235
		inode->i_flags |= S_APPEND;
D
Dave Chinner 已提交
1236
	if (flags & XFS_DIFLAG_SYNC)
1237
		inode->i_flags |= S_SYNC;
D
Dave Chinner 已提交
1238
	if (flags & XFS_DIFLAG_NOATIME)
1239
		inode->i_flags |= S_NOATIME;
1240
	if (xfs_inode_supports_dax(ip))
D
Dave Chinner 已提交
1241
		inode->i_flags |= S_DAX;
1242 1243 1244
}

/*
1245
 * Initialize the Linux inode.
1246
 *
1247 1248 1249 1250
 * 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.
1251 1252 1253 1254 1255
 */
void
xfs_setup_inode(
	struct xfs_inode	*ip)
{
1256
	struct inode		*inode = &ip->i_vnode;
1257
	gfp_t			gfp_mask;
1258 1259

	inode->i_ino = ip->i_ino;
C
Christoph Hellwig 已提交
1260
	inode->i_state = I_NEW;
1261 1262

	inode_sb_list_add(inode);
C
Christoph Hellwig 已提交
1263 1264
	/* make the inode look hashed for the writeback code */
	hlist_add_fake(&inode->i_hash);
1265

1266 1267
	inode->i_uid    = xfs_uid_to_kuid(ip->i_d.di_uid);
	inode->i_gid    = xfs_gid_to_kgid(ip->i_d.di_gid);
1268 1269 1270 1271

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

1272
	if (S_ISDIR(inode->i_mode)) {
1273
		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_dir_ilock_class);
1274
		ip->d_ops = ip->i_mount->m_dir_inode_ops;
1275 1276 1277
	} else {
		ip->d_ops = ip->i_mount->m_nondir_inode_ops;
		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_nondir_ilock_class);
1278 1279
	}

1280 1281 1282 1283 1284 1285 1286 1287
	/*
	 * 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)));

1288 1289 1290 1291 1292 1293
	/*
	 * 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);
1294
		cache_no_acl(inode);
1295
	}
1296
}
1297 1298 1299 1300 1301 1302 1303

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

1304 1305 1306 1307
	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &xfs_inode_operations;
		inode->i_fop = &xfs_file_operations;
D
Dan Williams 已提交
1308 1309 1310 1311
		if (IS_DAX(inode))
			inode->i_mapping->a_ops = &xfs_dax_aops;
		else
			inode->i_mapping->a_ops = &xfs_address_space_operations;
1312 1313 1314 1315 1316 1317 1318 1319 1320
		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:
1321 1322 1323 1324
		if (ip->i_df.if_flags & XFS_IFINLINE)
			inode->i_op = &xfs_inline_symlink_inode_operations;
		else
			inode->i_op = &xfs_symlink_inode_operations;
1325 1326 1327 1328 1329 1330 1331
		break;
	default:
		inode->i_op = &xfs_inode_operations;
		init_special_inode(inode, inode->i_mode, inode->i_rdev);
		break;
	}
}