xfs_iops.c 34.1 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 25 26
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_mount.h"
27
#include "xfs_da_format.h"
L
Linus Torvalds 已提交
28 29
#include "xfs_inode.h"
#include "xfs_bmap.h"
D
Dave Chinner 已提交
30
#include "xfs_bmap_util.h"
31 32
#include "xfs_acl.h"
#include "xfs_quota.h"
L
Linus Torvalds 已提交
33 34
#include "xfs_error.h"
#include "xfs_attr.h"
35
#include "xfs_trans.h"
C
Christoph Hellwig 已提交
36
#include "xfs_trace.h"
37
#include "xfs_icache.h"
D
Dave Chinner 已提交
38
#include "xfs_symlink.h"
39 40
#include "xfs_da_btree.h"
#include "xfs_dir2_priv.h"
41
#include "xfs_dinode.h"
Z
Zhi Yong Wu 已提交
42
#include "xfs_trans_space.h"
L
Linus Torvalds 已提交
43

44
#include <linux/capability.h>
L
Linus Torvalds 已提交
45 46
#include <linux/xattr.h>
#include <linux/namei.h>
47
#include <linux/posix_acl.h>
48
#include <linux/security.h>
E
Eric Sandeen 已提交
49
#include <linux/fiemap.h>
50
#include <linux/slab.h>
L
Linus Torvalds 已提交
51

52 53 54 55 56 57 58 59 60 61 62 63
/*
 * 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;

64 65 66 67 68
static int
xfs_initxattrs(
	struct inode		*inode,
	const struct xattr	*xattr_array,
	void			*fs_info)
69
{
70 71 72
	const struct xattr	*xattr;
	struct xfs_inode	*ip = XFS_I(inode);
	int			error = 0;
73 74

	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
D
Dave Chinner 已提交
75
		error = xfs_attr_set(ip, xattr->name, xattr->value,
76
				      xattr->value_len, ATTR_SECURE);
77 78 79 80 81 82
		if (error < 0)
			break;
	}
	return error;
}

83 84 85 86 87 88
/*
 * 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).
 */
89

90
STATIC int
91
xfs_init_security(
92
	struct inode	*inode,
93 94
	struct inode	*dir,
	const struct qstr *qstr)
95
{
D
Dave Chinner 已提交
96
	return security_inode_init_security(inode, dir, qstr,
97
					     &xfs_initxattrs, NULL);
98 99
}

100 101 102
static void
xfs_dentry_to_name(
	struct xfs_name	*namep,
103 104
	struct dentry	*dentry,
	int		mode)
105 106 107
{
	namep->name = dentry->d_name.name;
	namep->len = dentry->d_name.len;
108
	namep->type = xfs_mode_to_ftype[(mode & S_IFMT) >> S_SHIFT];
109 110
}

111
STATIC void
112
xfs_cleanup_inode(
113
	struct inode	*dir,
114
	struct inode	*inode,
115
	struct dentry	*dentry)
116
{
117
	struct xfs_name	teardown;
118 119

	/* Oh, the horror.
120
	 * If we can't add the ACL or we fail in
121
	 * xfs_init_security we must back out.
122 123
	 * ENOSPC can hit here, among other things.
	 */
124
	xfs_dentry_to_name(&teardown, dentry, 0);
125

126
	xfs_remove(XFS_I(dir), &teardown, XFS_I(inode));
127 128
}

L
Linus Torvalds 已提交
129
STATIC int
130
xfs_generic_create(
L
Linus Torvalds 已提交
131 132
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
133
	umode_t		mode,
134 135
	dev_t		rdev,
	bool		tmpfile)	/* unnamed file */
L
Linus Torvalds 已提交
136
{
C
Christoph Hellwig 已提交
137
	struct inode	*inode;
138
	struct xfs_inode *ip = NULL;
139
	struct posix_acl *default_acl, *acl;
140
	struct xfs_name	name;
L
Linus Torvalds 已提交
141 142 143 144 145 146
	int		error;

	/*
	 * Irix uses Missed'em'V split, but doesn't want to see
	 * the upper 5 bits of (14bit) major.
	 */
147 148 149 150 151 152 153
	if (S_ISCHR(mode) || S_ISBLK(mode)) {
		if (unlikely(!sysv_valid_dev(rdev) || MAJOR(rdev) & ~0x1ff))
			return -EINVAL;
		rdev = sysv_encode_dev(rdev);
	} else {
		rdev = 0;
	}
L
Linus Torvalds 已提交
154

155 156 157
	error = posix_acl_create(dir, &mode, &default_acl, &acl);
	if (error)
		return error;
L
Linus Torvalds 已提交
158

159 160 161 162 163 164
	if (!tmpfile) {
		xfs_dentry_to_name(&name, dentry, mode);
		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
165 166
	if (unlikely(error))
		goto out_free_acl;
167

168
	inode = VFS_I(ip);
169

170
	error = xfs_init_security(inode, dir, &dentry->d_name);
C
Christoph Hellwig 已提交
171 172 173
	if (unlikely(error))
		goto out_cleanup_inode;

174
#ifdef CONFIG_XFS_POSIX_ACL
C
Christoph Hellwig 已提交
175
	if (default_acl) {
D
Dave Chinner 已提交
176
		error = xfs_set_acl(inode, default_acl, ACL_TYPE_DEFAULT);
177
		if (error)
C
Christoph Hellwig 已提交
178
			goto out_cleanup_inode;
L
Linus Torvalds 已提交
179
	}
180
	if (acl) {
D
Dave Chinner 已提交
181
		error = xfs_set_acl(inode, acl, ACL_TYPE_ACCESS);
182 183 184 185
		if (error)
			goto out_cleanup_inode;
	}
#endif
L
Linus Torvalds 已提交
186

187 188 189 190 191
	if (tmpfile)
		d_tmpfile(dentry, inode);
	else
		d_instantiate(dentry, inode);

192 193 194 195 196
 out_free_acl:
	if (default_acl)
		posix_acl_release(default_acl);
	if (acl)
		posix_acl_release(acl);
D
Dave Chinner 已提交
197
	return error;
C
Christoph Hellwig 已提交
198 199

 out_cleanup_inode:
200 201 202
	if (!tmpfile)
		xfs_cleanup_inode(dir, inode, dentry);
	iput(inode);
203
	goto out_free_acl;
L
Linus Torvalds 已提交
204 205
}

206 207 208 209 210 211 212 213 214 215
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
216
STATIC int
217
xfs_vn_create(
L
Linus Torvalds 已提交
218 219
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
220
	umode_t		mode,
A
Al Viro 已提交
221
	bool		flags)
L
Linus Torvalds 已提交
222
{
223
	return xfs_vn_mknod(dir, dentry, mode, 0);
L
Linus Torvalds 已提交
224 225 226
}

STATIC int
227
xfs_vn_mkdir(
L
Linus Torvalds 已提交
228 229
	struct inode	*dir,
	struct dentry	*dentry,
230
	umode_t		mode)
L
Linus Torvalds 已提交
231
{
232
	return xfs_vn_mknod(dir, dentry, mode|S_IFDIR, 0);
L
Linus Torvalds 已提交
233 234 235
}

STATIC struct dentry *
236
xfs_vn_lookup(
L
Linus Torvalds 已提交
237 238
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
239
	unsigned int flags)
L
Linus Torvalds 已提交
240
{
241
	struct xfs_inode *cip;
242
	struct xfs_name	name;
L
Linus Torvalds 已提交
243 244 245 246 247
	int		error;

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

248
	xfs_dentry_to_name(&name, dentry, 0);
249
	error = xfs_lookup(XFS_I(dir), &name, &cip, NULL);
250
	if (unlikely(error)) {
D
Dave Chinner 已提交
251 252
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
L
Linus Torvalds 已提交
253 254 255 256
		d_add(dentry, NULL);
		return NULL;
	}

257
	return d_splice_alias(VFS_I(cip), dentry);
L
Linus Torvalds 已提交
258 259
}

260 261 262 263
STATIC struct dentry *
xfs_vn_ci_lookup(
	struct inode	*dir,
	struct dentry	*dentry,
A
Al Viro 已提交
264
	unsigned int flags)
265 266 267 268 269 270 271 272 273 274
{
	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);

275
	xfs_dentry_to_name(&xname, dentry, 0);
276 277
	error = xfs_lookup(XFS_I(dir), &xname, &ip, &ci_name);
	if (unlikely(error)) {
D
Dave Chinner 已提交
278 279
		if (unlikely(error != -ENOENT))
			return ERR_PTR(error);
280 281 282 283 284
		/*
		 * call d_add(dentry, NULL) here when d_drop_negative_children
		 * is called in xfs_vn_mknod (ie. allow negative dentries
		 * with CI filesystems).
		 */
285 286 287 288 289
		return NULL;
	}

	/* if exact match, just splice and exit */
	if (!ci_name.name)
290
		return d_splice_alias(VFS_I(ip), dentry);
291 292 293 294

	/* else case-insensitive match... */
	dname.name = ci_name.name;
	dname.len = ci_name.len;
295
	dentry = d_add_ci(dentry, VFS_I(ip), &dname);
296 297 298 299
	kmem_free(ci_name.name);
	return dentry;
}

L
Linus Torvalds 已提交
300
STATIC int
301
xfs_vn_link(
L
Linus Torvalds 已提交
302 303 304 305
	struct dentry	*old_dentry,
	struct inode	*dir,
	struct dentry	*dentry)
{
306
	struct inode	*inode = old_dentry->d_inode;
307
	struct xfs_name	name;
L
Linus Torvalds 已提交
308 309
	int		error;

310
	xfs_dentry_to_name(&name, dentry, inode->i_mode);
L
Linus Torvalds 已提交
311

312
	error = xfs_link(XFS_I(dir), XFS_I(inode), &name);
313
	if (unlikely(error))
D
Dave Chinner 已提交
314
		return error;
315

A
Al Viro 已提交
316
	ihold(inode);
317 318
	d_instantiate(dentry, inode);
	return 0;
L
Linus Torvalds 已提交
319 320 321
}

STATIC int
322
xfs_vn_unlink(
L
Linus Torvalds 已提交
323 324 325
	struct inode	*dir,
	struct dentry	*dentry)
{
326
	struct xfs_name	name;
L
Linus Torvalds 已提交
327 328
	int		error;

329
	xfs_dentry_to_name(&name, dentry, 0);
L
Linus Torvalds 已提交
330

D
Dave Chinner 已提交
331
	error = xfs_remove(XFS_I(dir), &name, XFS_I(dentry->d_inode));
332 333 334 335 336 337 338 339 340 341 342
	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
343 344 345
}

STATIC int
346
xfs_vn_symlink(
L
Linus Torvalds 已提交
347 348 349 350
	struct inode	*dir,
	struct dentry	*dentry,
	const char	*symname)
{
351 352
	struct inode	*inode;
	struct xfs_inode *cip = NULL;
353
	struct xfs_name	name;
L
Linus Torvalds 已提交
354
	int		error;
A
Al Viro 已提交
355
	umode_t		mode;
L
Linus Torvalds 已提交
356

357
	mode = S_IFLNK |
A
Al Viro 已提交
358
		(irix_symlink_mode ? 0777 & ~current_umask() : S_IRWXUGO);
359
	xfs_dentry_to_name(&name, dentry, mode);
L
Linus Torvalds 已提交
360

C
Christoph Hellwig 已提交
361
	error = xfs_symlink(XFS_I(dir), &name, symname, mode, &cip);
362 363 364
	if (unlikely(error))
		goto out;

365
	inode = VFS_I(cip);
366

367
	error = xfs_init_security(inode, dir, &dentry->d_name);
368 369 370 371 372 373 374
	if (unlikely(error))
		goto out_cleanup_inode;

	d_instantiate(dentry, inode);
	return 0;

 out_cleanup_inode:
375
	xfs_cleanup_inode(dir, inode, dentry);
376
	iput(inode);
377
 out:
D
Dave Chinner 已提交
378
	return error;
L
Linus Torvalds 已提交
379 380 381
}

STATIC int
382
xfs_vn_rename(
L
Linus Torvalds 已提交
383 384 385 386 387 388
	struct inode	*odir,
	struct dentry	*odentry,
	struct inode	*ndir,
	struct dentry	*ndentry)
{
	struct inode	*new_inode = ndentry->d_inode;
389 390
	struct xfs_name	oname;
	struct xfs_name	nname;
L
Linus Torvalds 已提交
391

392 393
	xfs_dentry_to_name(&oname, odentry, 0);
	xfs_dentry_to_name(&nname, ndentry, odentry->d_inode->i_mode);
394

D
Dave Chinner 已提交
395 396
	return xfs_rename(XFS_I(odir), &oname, XFS_I(odentry->d_inode),
			  XFS_I(ndir), &nname, new_inode ?
397
						XFS_I(new_inode) : NULL);
L
Linus Torvalds 已提交
398 399 400 401 402 403 404
}

/*
 * 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...
 */
405
STATIC void *
406
xfs_vn_follow_link(
L
Linus Torvalds 已提交
407 408 409 410
	struct dentry		*dentry,
	struct nameidata	*nd)
{
	char			*link;
411
	int			error = -ENOMEM;
L
Linus Torvalds 已提交
412

413
	link = kmalloc(MAXPATHLEN+1, GFP_KERNEL);
414 415
	if (!link)
		goto out_err;
L
Linus Torvalds 已提交
416

D
Dave Chinner 已提交
417
	error = xfs_readlink(XFS_I(dentry->d_inode), link);
418 419
	if (unlikely(error))
		goto out_kfree;
L
Linus Torvalds 已提交
420 421

	nd_set_link(nd, link);
422
	return NULL;
423 424 425 426 427 428

 out_kfree:
	kfree(link);
 out_err:
	nd_set_link(nd, ERR_PTR(error));
	return NULL;
L
Linus Torvalds 已提交
429 430 431
}

STATIC int
432
xfs_vn_getattr(
433 434 435
	struct vfsmount		*mnt,
	struct dentry		*dentry,
	struct kstat		*stat)
L
Linus Torvalds 已提交
436
{
437 438 439 440
	struct inode		*inode = dentry->d_inode;
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;

C
Christoph Hellwig 已提交
441
	trace_xfs_getattr(ip);
442 443

	if (XFS_FORCED_SHUTDOWN(mp))
E
Eric Sandeen 已提交
444
		return -EIO;
445 446 447 448 449

	stat->size = XFS_ISIZE(ip);
	stat->dev = inode->i_sb->s_dev;
	stat->mode = ip->i_d.di_mode;
	stat->nlink = ip->i_d.di_nlink;
450 451
	stat->uid = inode->i_uid;
	stat->gid = inode->i_gid;
452 453
	stat->ino = ip->i_ino;
	stat->atime = inode->i_atime;
454 455
	stat->mtime = inode->i_mtime;
	stat->ctime = inode->i_ctime;
456 457 458 459 460 461 462 463 464 465 466 467
	stat->blocks =
		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);


	switch (inode->i_mode & S_IFMT) {
	case S_IFBLK:
	case S_IFCHR:
		stat->blksize = BLKDEV_IOSIZE;
		stat->rdev = MKDEV(sysv_major(ip->i_df.if_u2.if_rdev) & 0x1ff,
				   sysv_minor(ip->i_df.if_u2.if_rdev));
		break;
	default:
468
		if (XFS_IS_REALTIME_INODE(ip)) {
469 470 471 472 473 474 475 476 477 478 479
			/*
			 * 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;
480
	}
481 482

	return 0;
L
Linus Torvalds 已提交
483 484
}

485 486 487 488 489
static void
xfs_setattr_mode(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
490 491
	struct inode		*inode = VFS_I(ip);
	umode_t			mode = iattr->ia_mode;
492 493 494 495 496 497 498 499 500 501

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

	ip->i_d.di_mode &= S_IFMT;
	ip->i_d.di_mode |= mode & ~S_IFMT;

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

C
Christoph Hellwig 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
static void
xfs_setattr_time(
	struct xfs_inode	*ip,
	struct iattr		*iattr)
{
	struct inode		*inode = VFS_I(ip);

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

	if (iattr->ia_valid & ATTR_ATIME) {
		inode->i_atime = iattr->ia_atime;
		ip->i_d.di_atime.t_sec = iattr->ia_atime.tv_sec;
		ip->i_d.di_atime.t_nsec = iattr->ia_atime.tv_nsec;
	}
	if (iattr->ia_valid & ATTR_CTIME) {
		inode->i_ctime = iattr->ia_ctime;
		ip->i_d.di_ctime.t_sec = iattr->ia_ctime.tv_sec;
		ip->i_d.di_ctime.t_nsec = iattr->ia_ctime.tv_nsec;
	}
	if (iattr->ia_valid & ATTR_MTIME) {
		inode->i_mtime = iattr->ia_mtime;
		ip->i_d.di_mtime.t_sec = iattr->ia_mtime.tv_sec;
		ip->i_d.di_mtime.t_nsec = iattr->ia_mtime.tv_nsec;
	}
}

C
Christoph Hellwig 已提交
528 529 530 531 532 533 534 535 536 537 538
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;
539 540
	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
541 542 543 544 545
	struct xfs_dquot	*udqp = NULL, *gdqp = NULL;
	struct xfs_dquot	*olddquot1 = NULL, *olddquot2 = NULL;

	trace_xfs_setattr(ip);

546 547 548
	/* If acls are being inherited, we already have this checked */
	if (!(flags & XFS_ATTR_NOACL)) {
		if (mp->m_flags & XFS_MOUNT_RDONLY)
D
Dave Chinner 已提交
549
			return -EROFS;
C
Christoph Hellwig 已提交
550

551
		if (XFS_FORCED_SHUTDOWN(mp))
D
Dave Chinner 已提交
552
			return -EIO;
C
Christoph Hellwig 已提交
553

D
Dave Chinner 已提交
554
		error = inode_change_ok(inode, iattr);
555
		if (error)
E
Eric Sandeen 已提交
556
			return error;
557
	}
C
Christoph Hellwig 已提交
558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575

	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 {
576
			uid = inode->i_uid;
C
Christoph Hellwig 已提交
577 578 579 580 581
		}
		if ((mask & ATTR_GID) && XFS_IS_GQUOTA_ON(mp)) {
			gid = iattr->ia_gid;
			qflags |= XFS_QMOPT_GQUOTA;
		}  else {
582
			gid = inode->i_gid;
C
Christoph Hellwig 已提交
583 584 585 586 587 588 589 590 591
		}

		/*
		 * 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);
592 593 594 595
		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
596 597 598 599 600
		if (error)
			return error;
	}

	tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
601
	error = xfs_trans_reserve(tp, &M_RES(mp)->tr_ichange, 0, 0);
C
Christoph Hellwig 已提交
602 603 604 605 606 607 608 609 610 611 612 613 614 615 616
	if (error)
		goto out_dqrele;

	xfs_ilock(ip, XFS_ILOCK_EXCL);

	/*
	 * 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.
		 */
617 618
		iuid = inode->i_uid;
		igid = inode->i_gid;
C
Christoph Hellwig 已提交
619 620 621 622 623 624 625 626
		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) &&
627 628
		    ((XFS_IS_UQUOTA_ON(mp) && !uid_eq(iuid, uid)) ||
		     (XFS_IS_GQUOTA_ON(mp) && !gid_eq(igid, gid)))) {
C
Christoph Hellwig 已提交
629 630
			ASSERT(tp);
			error = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
631
						NULL, capable(CAP_FOWNER) ?
C
Christoph Hellwig 已提交
632 633 634 635 636 637
						XFS_QMOPT_FORCE_RES : 0);
			if (error)	/* out of quota */
				goto out_trans_cancel;
		}
	}

638
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657

	/*
	 * 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()
		 */
		if ((ip->i_d.di_mode & (S_ISUID|S_ISGID)) &&
		    !capable(CAP_FSETID))
			ip->i_d.di_mode &= ~(S_ISUID|S_ISGID);

		/*
		 * Change the ownerships and register quota modifications
		 * in the transaction.
		 */
658
		if (!uid_eq(iuid, uid)) {
C
Christoph Hellwig 已提交
659 660 661 662 663 664
			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);
			}
665
			ip->i_d.di_uid = xfs_kuid_to_uid(uid);
C
Christoph Hellwig 已提交
666 667
			inode->i_uid = uid;
		}
668
		if (!gid_eq(igid, gid)) {
C
Christoph Hellwig 已提交
669
			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_GQUOTA_ON(mp)) {
670 671
				ASSERT(xfs_sb_version_has_pquotino(&mp->m_sb) ||
				       !XFS_IS_PQUOTA_ON(mp));
C
Christoph Hellwig 已提交
672 673 674 675 676
				ASSERT(mask & ATTR_GID);
				ASSERT(gdqp);
				olddquot2 = xfs_qm_vop_chown(tp, ip,
							&ip->i_gdquot, gdqp);
			}
677
			ip->i_d.di_gid = xfs_kgid_to_gid(gid);
C
Christoph Hellwig 已提交
678 679 680 681
			inode->i_gid = gid;
		}
	}

682
	if (mask & ATTR_MODE)
683
		xfs_setattr_mode(ip, iattr);
C
Christoph Hellwig 已提交
684 685
	if (mask & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	XFS_STATS_INC(xs_ig_attrchg);

	if (mp->m_flags & XFS_MOUNT_WSYNC)
		xfs_trans_set_sync(tp);
	error = xfs_trans_commit(tp, 0);

	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
706
		return error;
C
Christoph Hellwig 已提交
707 708 709 710 711 712 713 714 715

	/*
	 * 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
716
		error = posix_acl_chmod(inode, inode->i_mode);
C
Christoph Hellwig 已提交
717
		if (error)
E
Eric Sandeen 已提交
718
			return error;
C
Christoph Hellwig 已提交
719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
	}

	return 0;

out_trans_cancel:
	xfs_trans_cancel(tp, 0);
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
out_dqrele:
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	return error;
}

/*
 * Truncate file.  Must have write permission and not be a directory.
 */
int
xfs_setattr_size(
	struct xfs_inode	*ip,
738
	struct iattr		*iattr)
C
Christoph Hellwig 已提交
739 740 741
{
	struct xfs_mount	*mp = ip->i_mount;
	struct inode		*inode = VFS_I(ip);
742
	xfs_off_t		oldsize, newsize;
C
Christoph Hellwig 已提交
743 744
	struct xfs_trans	*tp;
	int			error;
745
	uint			lock_flags = 0;
C
Christoph Hellwig 已提交
746 747 748 749 750
	uint			commit_flags = 0;

	trace_xfs_setattr(ip);

	if (mp->m_flags & XFS_MOUNT_RDONLY)
D
Dave Chinner 已提交
751
		return -EROFS;
C
Christoph Hellwig 已提交
752 753

	if (XFS_FORCED_SHUTDOWN(mp))
D
Dave Chinner 已提交
754
		return -EIO;
C
Christoph Hellwig 已提交
755

D
Dave Chinner 已提交
756
	error = inode_change_ok(inode, iattr);
C
Christoph Hellwig 已提交
757
	if (error)
E
Eric Sandeen 已提交
758
		return error;
C
Christoph Hellwig 已提交
759

760
	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
C
Christoph Hellwig 已提交
761
	ASSERT(S_ISREG(ip->i_d.di_mode));
762 763
	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
764

765
	oldsize = inode->i_size;
766 767
	newsize = iattr->ia_size;

C
Christoph Hellwig 已提交
768 769 770
	/*
	 * Short circuit the truncate case for zero length files.
	 */
771
	if (newsize == 0 && oldsize == 0 && ip->i_d.di_nextents == 0) {
772
		if (!(iattr->ia_valid & (ATTR_CTIME|ATTR_MTIME)))
773
			return 0;
774 775 776 777 778 779

		/*
		 * 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
780 781 782 783 784
	}

	/*
	 * Make sure that the dquots are attached to the inode.
	 */
785
	error = xfs_qm_dqattach(ip, 0);
C
Christoph Hellwig 已提交
786
	if (error)
787
		return error;
C
Christoph Hellwig 已提交
788 789 790 791 792 793 794 795

	/*
	 * Now we can make the changes.  Before we join the inode to the
	 * transaction, take care of the part of the truncation that must be
	 * done without the inode lock.  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.
	 */
796
	if (newsize > oldsize) {
C
Christoph Hellwig 已提交
797 798 799 800 801 802
		/*
		 * Do the first part of growing a file: zero any data in the
		 * last block that is beyond the old EOF.  We need to do this
		 * before the inode is joined to the transaction to modify
		 * i_size.
		 */
803
		error = xfs_zero_eof(ip, newsize, oldsize);
C
Christoph Hellwig 已提交
804
		if (error)
805
			return error;
C
Christoph Hellwig 已提交
806 807 808 809 810 811 812 813 814 815 816 817 818 819
	}

	/*
	 * 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.
	 *
	 * Only flush from the on disk size to the smaller of the in memory
	 * file size or the new size as that's the range we really care about
	 * here and prevents waiting for other data not within the range we
	 * care about here.
	 */
820
	if (oldsize != ip->i_d.di_size && newsize > ip->i_d.di_size) {
D
Dave Chinner 已提交
821
		error = filemap_write_and_wait_range(VFS_I(ip)->i_mapping,
D
Dave Chinner 已提交
822
						      ip->i_d.di_size, newsize);
C
Christoph Hellwig 已提交
823
		if (error)
824
			return error;
C
Christoph Hellwig 已提交
825 826 827
	}

	/*
C
Christoph Hellwig 已提交
828
	 * Wait for all direct I/O to complete.
C
Christoph Hellwig 已提交
829
	 */
C
Christoph Hellwig 已提交
830
	inode_dio_wait(inode);
C
Christoph Hellwig 已提交
831

832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
	/*
	 * Do all the page cache truncate work outside the transaction context
	 * as the "lock" order is page lock->log space reservation.  i.e.
	 * locking pages inside the transaction can ABBA deadlock with
	 * writeback. We have to do the VFS inode size update before we truncate
	 * the pagecache, however, to avoid racing with page faults beyond the
	 * new EOF they are not serialised against truncate operations except by
	 * page locks and size updates.
	 *
	 * Hence we are in a situation where a truncate can fail with ENOMEM
	 * from xfs_trans_reserve(), but 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.
	 */
D
Dave Chinner 已提交
847
	error = block_truncate_page(inode->i_mapping, newsize, xfs_get_blocks);
C
Christoph Hellwig 已提交
848
	if (error)
849
		return error;
850
	truncate_setsize(inode, newsize);
C
Christoph Hellwig 已提交
851

852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881
	/*
	 * The "we can't serialise against page faults" pain gets worse.
	 *
	 * If the file is mapped then we have to clean the page at the old EOF
	 * when extending the file. Extending the file can expose changes the
	 * underlying page mapping (e.g. from beyond EOF to a hole or
	 * unwritten), and so on the next attempt to write to that page we need
	 * to remap it for write. i.e. we need .page_mkwrite() to be called.
	 * Hence we need to clean the page to clean the pte and so a new write
	 * fault will be triggered appropriately.
	 *
	 * If we do it before we change the inode size, then we can race with a
	 * page fault that maps the page with exactly the same problem. If we do
	 * it after we change the file size, then a new page fault can come in
	 * and allocate space before we've run the rest of the truncate
	 * transaction. That's kinda grotesque, but it's better than have data
	 * over a hole, and so that's the lesser evil that has been chosen here.
	 *
	 * The real solution, however, is to have some mechanism for locking out
	 * page faults while a truncate is in progress.
	 */
	if (newsize > oldsize && mapping_mapped(VFS_I(ip)->i_mapping)) {
		error = filemap_write_and_wait_range(
				VFS_I(ip)->i_mapping,
				round_down(oldsize, PAGE_CACHE_SIZE),
				round_up(oldsize, PAGE_CACHE_SIZE) - 1);
		if (error)
			return error;
	}

C
Christoph Hellwig 已提交
882
	tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_SIZE);
883
	error = xfs_trans_reserve(tp, &M_RES(mp)->tr_itruncate, 0, 0);
C
Christoph Hellwig 已提交
884 885 886 887 888 889
	if (error)
		goto out_trans_cancel;

	commit_flags = XFS_TRANS_RELEASE_LOG_RES;
	lock_flags |= XFS_ILOCK_EXCL;
	xfs_ilock(ip, XFS_ILOCK_EXCL);
890
	xfs_trans_ijoin(tp, ip, 0);
C
Christoph Hellwig 已提交
891 892 893 894 895 896 897 898 899 900 901

	/*
	 * 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.
	 */
902 903
	if (newsize != oldsize &&
	    !(iattr->ia_valid & (ATTR_CTIME | ATTR_MTIME))) {
C
Christoph Hellwig 已提交
904 905
		iattr->ia_ctime = iattr->ia_mtime =
			current_fs_time(inode->i_sb);
906
		iattr->ia_valid |= ATTR_CTIME | ATTR_MTIME;
C
Christoph Hellwig 已提交
907 908
	}

909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925
	/*
	 * 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
926 927 928 929 930 931 932 933 934 935 936
		if (error)
			goto out_trans_abort;

		/*
		 * 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);
937 938 939

		/* A truncate down always removes post-EOF blocks. */
		xfs_inode_clear_eofblocks_tag(ip);
C
Christoph Hellwig 已提交
940 941
	}

942
	if (iattr->ia_valid & ATTR_MODE)
943
		xfs_setattr_mode(ip, iattr);
944
	if (iattr->ia_valid & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
C
Christoph Hellwig 已提交
945
		xfs_setattr_time(ip, iattr);
C
Christoph Hellwig 已提交
946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966

	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	XFS_STATS_INC(xs_ig_attrchg);

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

	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
out_unlock:
	if (lock_flags)
		xfs_iunlock(ip, lock_flags);
	return error;

out_trans_abort:
	commit_flags |= XFS_TRANS_ABORT;
out_trans_cancel:
	xfs_trans_cancel(tp, commit_flags);
	goto out_unlock;
}

L
Linus Torvalds 已提交
967
STATIC int
968
xfs_vn_setattr(
969 970
	struct dentry		*dentry,
	struct iattr		*iattr)
L
Linus Torvalds 已提交
971
{
972 973 974 975 976 977 978 979 980 981 982
	struct xfs_inode	*ip = XFS_I(dentry->d_inode);
	int			error;

	if (iattr->ia_valid & ATTR_SIZE) {
		xfs_ilock(ip, XFS_IOLOCK_EXCL);
		error = xfs_setattr_size(ip, iattr);
		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
	} else {
		error = xfs_setattr_nonsize(ip, iattr, 0);
	}

D
Dave Chinner 已提交
983
	return error;
L
Linus Torvalds 已提交
984 985
}

C
Christoph Hellwig 已提交
986 987 988 989 990 991 992 993 994 995 996 997 998 999
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);

	tp = xfs_trans_alloc(mp, XFS_TRANS_FSYNC_TS);
1000
	error = xfs_trans_reserve(tp, &M_RES(mp)->tr_fsyncts, 0, 0);
C
Christoph Hellwig 已提交
1001 1002
	if (error) {
		xfs_trans_cancel(tp, 0);
D
Dave Chinner 已提交
1003
		return error;
C
Christoph Hellwig 已提交
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
	}

	xfs_ilock(ip, XFS_ILOCK_EXCL);
	if (flags & S_CTIME) {
		inode->i_ctime = *now;
		ip->i_d.di_ctime.t_sec = (__int32_t)now->tv_sec;
		ip->i_d.di_ctime.t_nsec = (__int32_t)now->tv_nsec;
	}
	if (flags & S_MTIME) {
		inode->i_mtime = *now;
		ip->i_d.di_mtime.t_sec = (__int32_t)now->tv_sec;
		ip->i_d.di_mtime.t_nsec = (__int32_t)now->tv_nsec;
	}
	if (flags & S_ATIME) {
		inode->i_atime = *now;
		ip->i_d.di_atime.t_sec = (__int32_t)now->tv_sec;
		ip->i_d.di_atime.t_nsec = (__int32_t)now->tv_nsec;
	}
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
	xfs_trans_log_inode(tp, ip, XFS_ILOG_TIMESTAMP);
D
Dave Chinner 已提交
1024
	return xfs_trans_commit(tp, 0);
C
Christoph Hellwig 已提交
1025 1026
}

E
Eric Sandeen 已提交
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
#define XFS_FIEMAP_FLAGS	(FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)

/*
 * Call fiemap helper to fill in user data.
 * Returns positive errors to xfs_getbmap.
 */
STATIC int
xfs_fiemap_format(
	void			**arg,
	struct getbmapx		*bmv,
	int			*full)
{
	int			error;
	struct fiemap_extent_info *fieinfo = *arg;
	u32			fiemap_flags = 0;
	u64			logical, physical, length;

	/* Do nothing for a hole */
	if (bmv->bmv_block == -1LL)
		return 0;

	logical = BBTOB(bmv->bmv_offset);
	physical = BBTOB(bmv->bmv_block);
	length = BBTOB(bmv->bmv_length);

	if (bmv->bmv_oflags & BMV_OF_PREALLOC)
		fiemap_flags |= FIEMAP_EXTENT_UNWRITTEN;
	else if (bmv->bmv_oflags & BMV_OF_DELALLOC) {
1055 1056
		fiemap_flags |= (FIEMAP_EXTENT_DELALLOC |
				 FIEMAP_EXTENT_UNKNOWN);
E
Eric Sandeen 已提交
1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068
		physical = 0;   /* no block yet */
	}
	if (bmv->bmv_oflags & BMV_OF_LAST)
		fiemap_flags |= FIEMAP_EXTENT_LAST;

	error = fiemap_fill_next_extent(fieinfo, logical, physical,
					length, fiemap_flags);
	if (error > 0) {
		error = 0;
		*full = 1;	/* user array now full */
	}

D
Dave Chinner 已提交
1069
	return error;
E
Eric Sandeen 已提交
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
}

STATIC int
xfs_vn_fiemap(
	struct inode		*inode,
	struct fiemap_extent_info *fieinfo,
	u64			start,
	u64			length)
{
	xfs_inode_t		*ip = XFS_I(inode);
	struct getbmapx		bm;
	int			error;

	error = fiemap_check_flags(fieinfo, XFS_FIEMAP_FLAGS);
	if (error)
		return error;

	/* Set up bmap header for xfs internal routine */
1088
	bm.bmv_offset = BTOBBT(start);
E
Eric Sandeen 已提交
1089 1090 1091 1092
	/* Special case for whole file */
	if (length == FIEMAP_MAX_OFFSET)
		bm.bmv_length = -1LL;
	else
1093
		bm.bmv_length = BTOBB(start + length) - bm.bmv_offset;
E
Eric Sandeen 已提交
1094

1095
	/* We add one because in getbmap world count includes the header */
T
Tao Ma 已提交
1096 1097 1098 1099
	bm.bmv_count = !fieinfo->fi_extents_max ? MAXEXTNUM :
					fieinfo->fi_extents_max + 1;
	bm.bmv_count = min_t(__s32, bm.bmv_count,
			     (PAGE_SIZE * 16 / sizeof(struct getbmapx)));
T
Tao Ma 已提交
1100
	bm.bmv_iflags = BMV_IF_PREALLOC | BMV_IF_NO_HOLES;
E
Eric Sandeen 已提交
1101 1102 1103 1104 1105 1106 1107
	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR)
		bm.bmv_iflags |= BMV_IF_ATTRFORK;
	if (!(fieinfo->fi_flags & FIEMAP_FLAG_SYNC))
		bm.bmv_iflags |= BMV_IF_DELALLOC;

	error = xfs_getbmap(ip, &bm, xfs_fiemap_format, fieinfo);
	if (error)
D
Dave Chinner 已提交
1108
		return error;
E
Eric Sandeen 已提交
1109 1110 1111 1112

	return 0;
}

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
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
1127 1128 1129
	.setxattr		= generic_setxattr,
	.getxattr		= generic_getxattr,
	.removexattr		= generic_removexattr,
1130
	.listxattr		= xfs_vn_listxattr,
E
Eric Sandeen 已提交
1131
	.fiemap			= xfs_vn_fiemap,
C
Christoph Hellwig 已提交
1132
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1133 1134
};

1135
static const struct inode_operations xfs_dir_inode_operations = {
1136 1137 1138 1139 1140 1141
	.create			= xfs_vn_create,
	.lookup			= xfs_vn_lookup,
	.link			= xfs_vn_link,
	.unlink			= xfs_vn_unlink,
	.symlink		= xfs_vn_symlink,
	.mkdir			= xfs_vn_mkdir,
1142 1143 1144 1145 1146 1147 1148
	/*
	 * 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,
1149 1150
	.mknod			= xfs_vn_mknod,
	.rename			= xfs_vn_rename,
1151
	.get_acl		= xfs_get_acl,
1152
	.set_acl		= xfs_set_acl,
1153 1154
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
1155 1156 1157
	.setxattr		= generic_setxattr,
	.getxattr		= generic_getxattr,
	.removexattr		= generic_removexattr,
1158
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1159
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1160
	.tmpfile		= xfs_vn_tmpfile,
L
Linus Torvalds 已提交
1161 1162
};

1163
static const struct inode_operations xfs_dir_ci_inode_operations = {
1164 1165 1166 1167 1168 1169
	.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,
1170 1171 1172 1173 1174 1175 1176
	/*
	 * 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,
1177 1178
	.mknod			= xfs_vn_mknod,
	.rename			= xfs_vn_rename,
1179
	.get_acl		= xfs_get_acl,
1180
	.set_acl		= xfs_set_acl,
1181 1182
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
1183 1184 1185
	.setxattr		= generic_setxattr,
	.getxattr		= generic_getxattr,
	.removexattr		= generic_removexattr,
1186
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1187
	.update_time		= xfs_vn_update_time,
Z
Zhi Yong Wu 已提交
1188
	.tmpfile		= xfs_vn_tmpfile,
1189 1190
};

1191
static const struct inode_operations xfs_symlink_inode_operations = {
L
Linus Torvalds 已提交
1192
	.readlink		= generic_readlink,
1193
	.follow_link		= xfs_vn_follow_link,
A
Al Viro 已提交
1194
	.put_link		= kfree_put_link,
1195 1196
	.getattr		= xfs_vn_getattr,
	.setattr		= xfs_vn_setattr,
1197 1198 1199
	.setxattr		= generic_setxattr,
	.getxattr		= generic_getxattr,
	.removexattr		= generic_removexattr,
1200
	.listxattr		= xfs_vn_listxattr,
C
Christoph Hellwig 已提交
1201
	.update_time		= xfs_vn_update_time,
L
Linus Torvalds 已提交
1202
};
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233

STATIC void
xfs_diflags_to_iflags(
	struct inode		*inode,
	struct xfs_inode	*ip)
{
	if (ip->i_d.di_flags & XFS_DIFLAG_IMMUTABLE)
		inode->i_flags |= S_IMMUTABLE;
	else
		inode->i_flags &= ~S_IMMUTABLE;
	if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
		inode->i_flags |= S_APPEND;
	else
		inode->i_flags &= ~S_APPEND;
	if (ip->i_d.di_flags & XFS_DIFLAG_SYNC)
		inode->i_flags |= S_SYNC;
	else
		inode->i_flags &= ~S_SYNC;
	if (ip->i_d.di_flags & XFS_DIFLAG_NOATIME)
		inode->i_flags |= S_NOATIME;
	else
		inode->i_flags &= ~S_NOATIME;
}

/*
 * Initialize the Linux inode, set up the operation vectors and
 * unlock the inode.
 *
 * 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.
1234 1235 1236 1237
 *
 * We are always called with an uninitialised linux inode here.
 * We need to initialise the necessary fields and take a reference
 * on it.
1238 1239 1240 1241 1242
 */
void
xfs_setup_inode(
	struct xfs_inode	*ip)
{
1243
	struct inode		*inode = &ip->i_vnode;
1244
	gfp_t			gfp_mask;
1245 1246

	inode->i_ino = ip->i_ino;
C
Christoph Hellwig 已提交
1247
	inode->i_state = I_NEW;
1248 1249

	inode_sb_list_add(inode);
C
Christoph Hellwig 已提交
1250 1251
	/* make the inode look hashed for the writeback code */
	hlist_add_fake(&inode->i_hash);
1252 1253

	inode->i_mode	= ip->i_d.di_mode;
M
Miklos Szeredi 已提交
1254
	set_nlink(inode, ip->i_d.di_nlink);
1255 1256
	inode->i_uid    = xfs_uid_to_kuid(ip->i_d.di_uid);
	inode->i_gid    = xfs_gid_to_kgid(ip->i_d.di_gid);
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279

	switch (inode->i_mode & S_IFMT) {
	case S_IFBLK:
	case S_IFCHR:
		inode->i_rdev =
			MKDEV(sysv_major(ip->i_df.if_u2.if_rdev) & 0x1ff,
			      sysv_minor(ip->i_df.if_u2.if_rdev));
		break;
	default:
		inode->i_rdev = 0;
		break;
	}

	inode->i_generation = ip->i_d.di_gen;
	i_size_write(inode, ip->i_d.di_size);
	inode->i_atime.tv_sec	= ip->i_d.di_atime.t_sec;
	inode->i_atime.tv_nsec	= ip->i_d.di_atime.t_nsec;
	inode->i_mtime.tv_sec	= ip->i_d.di_mtime.t_sec;
	inode->i_mtime.tv_nsec	= ip->i_d.di_mtime.t_nsec;
	inode->i_ctime.tv_sec	= ip->i_d.di_ctime.t_sec;
	inode->i_ctime.tv_nsec	= ip->i_d.di_ctime.t_nsec;
	xfs_diflags_to_iflags(inode, ip);

D
Dave Chinner 已提交
1280
	ip->d_ops = ip->i_mount->m_nondir_inode_ops;
1281
	lockdep_set_class(&ip->i_lock.mr_lock, &xfs_nondir_ilock_class);
1282 1283 1284 1285 1286 1287 1288
	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:
1289
		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_dir_ilock_class);
1290 1291 1292 1293 1294
		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;
1295
		ip->d_ops = ip->i_mount->m_dir_inode_ops;
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307
		break;
	case S_IFLNK:
		inode->i_op = &xfs_symlink_inode_operations;
		if (!(ip->i_df.if_flags & XFS_IFINLINE))
			inode->i_mapping->a_ops = &xfs_address_space_operations;
		break;
	default:
		inode->i_op = &xfs_inode_operations;
		init_special_inode(inode, inode->i_mode, inode->i_rdev);
		break;
	}

1308 1309 1310 1311 1312 1313 1314 1315
	/*
	 * 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)));

1316 1317 1318 1319 1320 1321
	/*
	 * 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);
1322
		cache_no_acl(inode);
1323
	}
1324

1325 1326 1327 1328 1329
	xfs_iflags_clear(ip, XFS_INEW);
	barrier();

	unlock_new_inode(inode);
}