inode.c 62.1 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
L
Linus Torvalds 已提交
2 3 4 5 6 7 8
/*
 *  linux/fs/nfs/inode.c
 *
 *  Copyright (C) 1992  Rick Sladkey
 *
 *  nfs inode and superblock handling functions
 *
A
Alan Cox 已提交
9
 *  Modularised by Alan Cox <alan@lxorguk.ukuu.org.uk>, while hacking some
L
Linus Torvalds 已提交
10 11 12 13 14 15 16 17 18
 *  experimental NFS changes. Modularisation taken straight from SYS5 fs.
 *
 *  Change to nfs_read_super() to permit NFS mounts to multi-homed hosts.
 *  J.S.Peatfield@damtp.cam.ac.uk
 *
 */

#include <linux/module.h>
#include <linux/init.h>
19
#include <linux/sched/signal.h>
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27 28
#include <linux/time.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/errno.h>
#include <linux/unistd.h>
#include <linux/sunrpc/clnt.h>
#include <linux/sunrpc/stats.h>
29
#include <linux/sunrpc/metrics.h>
L
Linus Torvalds 已提交
30 31 32 33 34 35 36
#include <linux/nfs_fs.h>
#include <linux/nfs_mount.h>
#include <linux/nfs4_mount.h>
#include <linux/lockd/bind.h>
#include <linux/seq_file.h>
#include <linux/mount.h>
#include <linux/vfs.h>
37 38
#include <linux/inet.h>
#include <linux/nfs_xdr.h>
39
#include <linux/slab.h>
40
#include <linux/compat.h>
41
#include <linux/freezer.h>
42
#include <linux/uaccess.h>
J
Jeff Layton 已提交
43
#include <linux/iversion.h>
L
Linus Torvalds 已提交
44

45
#include "nfs4_fs.h"
46
#include "callback.h"
L
Linus Torvalds 已提交
47
#include "delegation.h"
48
#include "iostat.h"
D
David Howells 已提交
49
#include "internal.h"
50
#include "fscache.h"
51
#include "pnfs.h"
52
#include "nfs.h"
53
#include "netns.h"
54
#include "sysfs.h"
L
Linus Torvalds 已提交
55

56 57
#include "nfstrace.h"

L
Linus Torvalds 已提交
58 59
#define NFSDBG_FACILITY		NFSDBG_VFS

60 61 62
#define NFS_64_BIT_INODE_NUMBERS_ENABLED	1

/* Default is to see 64-bit inode numbers */
63
static bool enable_ino64 = NFS_64_BIT_INODE_NUMBERS_ENABLED;
64

L
Linus Torvalds 已提交
65
static void nfs_invalidate_inode(struct inode *);
66
static int nfs_update_inode(struct inode *, struct nfs_fattr *);
L
Linus Torvalds 已提交
67

68
static struct kmem_cache * nfs_inode_cachep;
69

L
Linus Torvalds 已提交
70 71 72 73 74 75
static inline unsigned long
nfs_fattr_to_ino_t(struct nfs_fattr *fattr)
{
	return nfs_fileid_to_ino_t(fattr->fileid);
}

76
static int nfs_wait_killable(int mode)
77
{
78
	freezable_schedule_unsafe();
79 80
	if (signal_pending_state(mode, current))
		return -ERESTARTSYS;
81 82
	return 0;
}
83 84 85 86 87

int nfs_wait_bit_killable(struct wait_bit_key *key, int mode)
{
	return nfs_wait_killable(mode);
}
B
Bryan Schumaker 已提交
88
EXPORT_SYMBOL_GPL(nfs_wait_bit_killable);
89

90 91 92 93 94 95 96 97 98
/**
 * nfs_compat_user_ino64 - returns the user-visible inode number
 * @fileid: 64-bit fileid
 *
 * This function returns a 32-bit inode number if the boot parameter
 * nfs.enable_ino64 is zero.
 */
u64 nfs_compat_user_ino64(u64 fileid)
{
99 100 101 102 103
#ifdef CONFIG_COMPAT
	compat_ulong_t ino;
#else	
	unsigned long ino;
#endif
104 105 106 107 108 109 110 111 112

	if (enable_ino64)
		return fileid;
	ino = fileid;
	if (sizeof(ino) < sizeof(fileid))
		ino ^= fileid >> (sizeof(fileid)-sizeof(ino)) * 8;
	return ino;
}

113 114 115 116 117 118
int nfs_drop_inode(struct inode *inode)
{
	return NFS_STALE(inode) || generic_drop_inode(inode);
}
EXPORT_SYMBOL_GPL(nfs_drop_inode);

119
void nfs_clear_inode(struct inode *inode)
L
Linus Torvalds 已提交
120
{
T
Trond Myklebust 已提交
121 122 123
	/*
	 * The following should never happen...
	 */
124 125
	WARN_ON_ONCE(nfs_have_writebacks(inode));
	WARN_ON_ONCE(!list_empty(&NFS_I(inode)->open_files));
126
	nfs_zap_acl_cache(inode);
127
	nfs_access_zap_cache(inode);
128
	nfs_fscache_clear_inode(inode);
L
Linus Torvalds 已提交
129
}
B
Bryan Schumaker 已提交
130
EXPORT_SYMBOL_GPL(nfs_clear_inode);
L
Linus Torvalds 已提交
131

132 133
void nfs_evict_inode(struct inode *inode)
{
134
	truncate_inode_pages_final(&inode->i_data);
135
	clear_inode(inode);
136 137 138
	nfs_clear_inode(inode);
}

139
int nfs_sync_inode(struct inode *inode)
140
{
141
	inode_dio_wait(inode);
142 143
	return nfs_wb_all(inode);
}
144
EXPORT_SYMBOL_GPL(nfs_sync_inode);
145

T
Trond Myklebust 已提交
146 147
/**
 * nfs_sync_mapping - helper to flush all mmapped dirty data to disk
148
 * @mapping: pointer to struct address_space
T
Trond Myklebust 已提交
149 150 151
 */
int nfs_sync_mapping(struct address_space *mapping)
{
T
Trond Myklebust 已提交
152
	int ret = 0;
T
Trond Myklebust 已提交
153

T
Trond Myklebust 已提交
154 155 156 157
	if (mapping->nrpages != 0) {
		unmap_mapping_range(mapping, 0, 0, 0);
		ret = nfs_wb_all(mapping->host);
	}
T
Trond Myklebust 已提交
158 159 160
	return ret;
}

161 162 163 164 165 166 167
static int nfs_attribute_timeout(struct inode *inode)
{
	struct nfs_inode *nfsi = NFS_I(inode);

	return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
}

168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
static bool nfs_check_cache_invalid_delegated(struct inode *inode, unsigned long flags)
{
	unsigned long cache_validity = READ_ONCE(NFS_I(inode)->cache_validity);

	/* Special case for the pagecache or access cache */
	if (flags == NFS_INO_REVAL_PAGECACHE &&
	    !(cache_validity & NFS_INO_REVAL_FORCED))
		return false;
	return (cache_validity & flags) != 0;
}

static bool nfs_check_cache_invalid_not_delegated(struct inode *inode, unsigned long flags)
{
	unsigned long cache_validity = READ_ONCE(NFS_I(inode)->cache_validity);

	if ((cache_validity & flags) != 0)
		return true;
	if (nfs_attribute_timeout(inode))
		return true;
	return false;
}

bool nfs_check_cache_invalid(struct inode *inode, unsigned long flags)
{
	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
		return nfs_check_cache_invalid_delegated(inode, flags);

	return nfs_check_cache_invalid_not_delegated(inode, flags);
}

198 199 200
static void nfs_set_cache_invalid(struct inode *inode, unsigned long flags)
{
	struct nfs_inode *nfsi = NFS_I(inode);
201 202 203 204 205 206 207 208 209
	bool have_delegation = NFS_PROTO(inode)->have_delegation(inode, FMODE_READ);

	if (have_delegation) {
		if (!(flags & NFS_INO_REVAL_FORCED))
			flags &= ~NFS_INO_INVALID_OTHER;
		flags &= ~(NFS_INO_INVALID_CHANGE
				| NFS_INO_INVALID_SIZE
				| NFS_INO_REVAL_PAGECACHE);
	}
210 211

	if (inode->i_mapping->nrpages == 0)
212
		flags &= ~(NFS_INO_INVALID_DATA|NFS_INO_DATA_INVAL_DEFER);
213 214 215 216 217
	nfsi->cache_validity |= flags;
	if (flags & NFS_INO_INVALID_DATA)
		nfs_fscache_invalidate(inode);
}

L
Linus Torvalds 已提交
218 219 220
/*
 * Invalidate the local caches
 */
221
static void nfs_zap_caches_locked(struct inode *inode)
L
Linus Torvalds 已提交
222 223 224 225
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int mode = inode->i_mode;

C
Chuck Lever 已提交
226 227
	nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE);

228 229
	nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
	nfsi->attrtimeo_timestamp = jiffies;
L
Linus Torvalds 已提交
230

231
	memset(NFS_I(inode)->cookieverf, 0, sizeof(NFS_I(inode)->cookieverf));
D
David Howells 已提交
232
	if (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)) {
233
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
234 235 236
					| NFS_INO_INVALID_DATA
					| NFS_INO_INVALID_ACCESS
					| NFS_INO_INVALID_ACL
237
					| NFS_INO_REVAL_PAGECACHE);
238
	} else
239
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
240 241
					| NFS_INO_INVALID_ACCESS
					| NFS_INO_INVALID_ACL
242
					| NFS_INO_REVAL_PAGECACHE);
243
	nfs_zap_label_cache_locked(nfsi);
244
}
245

246 247 248 249
void nfs_zap_caches(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	nfs_zap_caches_locked(inode);
250
	spin_unlock(&inode->i_lock);
251 252
}

253 254 255 256
void nfs_zap_mapping(struct inode *inode, struct address_space *mapping)
{
	if (mapping->nrpages != 0) {
		spin_lock(&inode->i_lock);
257
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
258 259 260 261
		spin_unlock(&inode->i_lock);
	}
}

262
void nfs_zap_acl_cache(struct inode *inode)
263 264 265 266 267 268
{
	void (*clear_acl_cache)(struct inode *);

	clear_acl_cache = NFS_PROTO(inode)->clear_acl_cache;
	if (clear_acl_cache != NULL)
		clear_acl_cache(inode);
269
	spin_lock(&inode->i_lock);
270
	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_ACL;
271
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
272
}
B
Bryan Schumaker 已提交
273
EXPORT_SYMBOL_GPL(nfs_zap_acl_cache);
L
Linus Torvalds 已提交
274

275 276 277
void nfs_invalidate_atime(struct inode *inode)
{
	spin_lock(&inode->i_lock);
278
	nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME);
279 280
	spin_unlock(&inode->i_lock);
}
B
Bryan Schumaker 已提交
281
EXPORT_SYMBOL_GPL(nfs_invalidate_atime);
282

L
Linus Torvalds 已提交
283
/*
284 285
 * Invalidate, but do not unhash, the inode.
 * NB: must be called with inode->i_lock held!
L
Linus Torvalds 已提交
286
 */
287
static void nfs_invalidate_inode(struct inode *inode)
L
Linus Torvalds 已提交
288
{
B
Benny Halevy 已提交
289
	set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
290
	nfs_zap_caches_locked(inode);
L
Linus Torvalds 已提交
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
}

struct nfs_find_desc {
	struct nfs_fh		*fh;
	struct nfs_fattr	*fattr;
};

/*
 * In NFSv3 we can have 64bit inode numbers. In order to support
 * this, and re-exported directories (also seen in NFSv2)
 * we are forced to allow 2 different inodes to have the same
 * i_ino.
 */
static int
nfs_find_actor(struct inode *inode, void *opaque)
{
	struct nfs_find_desc	*desc = (struct nfs_find_desc *)opaque;
	struct nfs_fh		*fh = desc->fh;
	struct nfs_fattr	*fattr = desc->fattr;

	if (NFS_FILEID(inode) != fattr->fileid)
		return 0;
313 314
	if ((S_IFMT & inode->i_mode) != (S_IFMT & fattr->mode))
		return 0;
L
Linus Torvalds 已提交
315 316 317 318 319 320 321 322 323 324 325 326 327
	if (nfs_compare_fh(NFS_FH(inode), fh))
		return 0;
	if (is_bad_inode(inode) || NFS_STALE(inode))
		return 0;
	return 1;
}

static int
nfs_init_locked(struct inode *inode, void *opaque)
{
	struct nfs_find_desc	*desc = (struct nfs_find_desc *)opaque;
	struct nfs_fattr	*fattr = desc->fattr;

328
	set_nfs_fileid(inode, fattr->fileid);
329
	inode->i_mode = fattr->mode;
L
Linus Torvalds 已提交
330 331 332 333
	nfs_copy_fh(NFS_FH(inode), desc->fh);
	return 0;
}

334
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
335 336 337 338 339 340 341
static void nfs_clear_label_invalid(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_LABEL;
	spin_unlock(&inode->i_lock);
}

342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
					struct nfs4_label *label)
{
	int error;

	if (label == NULL)
		return;

	if ((fattr->valid & NFS_ATTR_FATTR_V4_SECURITY_LABEL) && inode->i_security) {
		error = security_inode_notifysecctx(inode, label->label,
				label->len);
		if (error)
			printk(KERN_ERR "%s() %s %d "
					"security_inode_notifysecctx() %d\n",
					__func__,
					(char *)label->label,
					label->len, error);
359
		nfs_clear_label_invalid(inode);
360 361 362
	}
}

363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
struct nfs4_label *nfs4_label_alloc(struct nfs_server *server, gfp_t flags)
{
	struct nfs4_label *label = NULL;
	int minor_version = server->nfs_client->cl_minorversion;

	if (minor_version < 2)
		return label;

	if (!(server->caps & NFS_CAP_SECURITY_LABEL))
		return label;

	label = kzalloc(sizeof(struct nfs4_label), flags);
	if (label == NULL)
		return ERR_PTR(-ENOMEM);

	label->label = kzalloc(NFS4_MAXLABELLEN, flags);
	if (label->label == NULL) {
		kfree(label);
		return ERR_PTR(-ENOMEM);
	}
	label->len = NFS4_MAXLABELLEN;

	return label;
}
EXPORT_SYMBOL_GPL(nfs4_label_alloc);
388
#else
389
void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
390 391 392
					struct nfs4_label *label)
{
}
393
#endif
394
EXPORT_SYMBOL_GPL(nfs_setsecurity);
395

P
Peng Tao 已提交
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
/* Search for inode identified by fh, fileid and i_mode in inode cache. */
struct inode *
nfs_ilookup(struct super_block *sb, struct nfs_fattr *fattr, struct nfs_fh *fh)
{
	struct nfs_find_desc desc = {
		.fh	= fh,
		.fattr	= fattr,
	};
	struct inode *inode;
	unsigned long hash;

	if (!(fattr->valid & NFS_ATTR_FATTR_FILEID) ||
	    !(fattr->valid & NFS_ATTR_FATTR_TYPE))
		return NULL;

	hash = nfs_fattr_to_ino_t(fattr);
	inode = ilookup5(sb, hash, nfs_find_actor, &desc);

	dprintk("%s: returning %p\n", __func__, inode);
	return inode;
}

L
Linus Torvalds 已提交
418 419 420 421 422
/*
 * This is our front-end to iget that looks up inodes by file handle
 * instead of inode number.
 */
struct inode *
423
nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
424 425 426 427 428
{
	struct nfs_find_desc desc = {
		.fh	= fh,
		.fattr	= fattr
	};
429
	struct inode *inode = ERR_PTR(-ENOENT);
L
Linus Torvalds 已提交
430 431
	unsigned long hash;

432 433
	nfs_attr_check_mountpoint(sb, fattr);

434 435 436
	if (nfs_attr_use_mounted_on_fileid(fattr))
		fattr->fileid = fattr->mounted_on_fileid;
	else if ((fattr->valid & NFS_ATTR_FATTR_FILEID) == 0)
L
Linus Torvalds 已提交
437
		goto out_no_inode;
438
	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) == 0)
L
Linus Torvalds 已提交
439 440 441 442
		goto out_no_inode;

	hash = nfs_fattr_to_ino_t(fattr);

443 444 445
	inode = iget5_locked(sb, hash, nfs_find_actor, nfs_init_locked, &desc);
	if (inode == NULL) {
		inode = ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
446
		goto out_no_inode;
447
	}
L
Linus Torvalds 已提交
448 449 450

	if (inode->i_state & I_NEW) {
		struct nfs_inode *nfsi = NFS_I(inode);
451
		unsigned long now = jiffies;
L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459

		/* We set i_ino for the few things that still rely on it,
		 * such as stat(2) */
		inode->i_ino = hash;

		/* We can't support update_atime(), since the server will reset it */
		inode->i_flags |= S_NOATIME|S_NOCMTIME;
		inode->i_mode = fattr->mode;
460
		nfsi->cache_validity = 0;
461 462
		if ((fattr->valid & NFS_ATTR_FATTR_MODE) == 0
				&& nfs_server_capable(inode, NFS_CAP_MODE))
463
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER);
L
Linus Torvalds 已提交
464 465 466
		/* Why so? Because we want revalidate for devices/FIFOs, and
		 * that's precisely what we have in nfs_file_inode_operations.
		 */
467
		inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->file_inode_ops;
L
Linus Torvalds 已提交
468
		if (S_ISREG(inode->i_mode)) {
469
			inode->i_fop = NFS_SB(sb)->nfs_client->rpc_ops->file_ops;
L
Linus Torvalds 已提交
470 471
			inode->i_data.a_ops = &nfs_file_aops;
		} else if (S_ISDIR(inode->i_mode)) {
472
			inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->dir_inode_ops;
L
Linus Torvalds 已提交
473
			inode->i_fop = &nfs_dir_operations;
474
			inode->i_data.a_ops = &nfs_dir_aops;
475
			/* Deal with crossing mountpoints */
476 477
			if (fattr->valid & NFS_ATTR_FATTR_MOUNTPOINT ||
					fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) {
M
Manoj Naik 已提交
478 479 480 481
				if (fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL)
					inode->i_op = &nfs_referral_inode_operations;
				else
					inode->i_op = &nfs_mountpoint_inode_operations;
482
				inode->i_fop = NULL;
483
				inode->i_flags |= S_AUTOMOUNT;
484
			}
485
		} else if (S_ISLNK(inode->i_mode)) {
L
Linus Torvalds 已提交
486
			inode->i_op = &nfs_symlink_inode_operations;
487 488
			inode_nohighmem(inode);
		} else
L
Linus Torvalds 已提交
489 490
			init_special_inode(inode, inode->i_mode, fattr->rdev);

491 492 493
		memset(&inode->i_atime, 0, sizeof(inode->i_atime));
		memset(&inode->i_mtime, 0, sizeof(inode->i_mtime));
		memset(&inode->i_ctime, 0, sizeof(inode->i_ctime));
J
Jeff Layton 已提交
494
		inode_set_iversion_raw(inode, 0);
495
		inode->i_size = 0;
496
		clear_nlink(inode);
497 498
		inode->i_uid = make_kuid(&init_user_ns, -2);
		inode->i_gid = make_kgid(&init_user_ns, -2);
499 500
		inode->i_blocks = 0;
		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
501 502
		nfsi->write_io = 0;
		nfsi->read_io = 0;
503

504
		nfsi->read_cache_jiffies = fattr->time_start;
505
		nfsi->attr_gencount = fattr->gencount;
506
		if (fattr->valid & NFS_ATTR_FATTR_ATIME)
507
			inode->i_atime = fattr->atime;
508
		else if (nfs_server_capable(inode, NFS_CAP_ATIME))
509
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME);
510
		if (fattr->valid & NFS_ATTR_FATTR_MTIME)
511
			inode->i_mtime = fattr->mtime;
512
		else if (nfs_server_capable(inode, NFS_CAP_MTIME))
513
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME);
514
		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
515
			inode->i_ctime = fattr->ctime;
516
		else if (nfs_server_capable(inode, NFS_CAP_CTIME))
517
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_CTIME);
518
		if (fattr->valid & NFS_ATTR_FATTR_CHANGE)
J
Jeff Layton 已提交
519
			inode_set_iversion_raw(inode, fattr->change_attr);
520
		else
521
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE);
522 523
		if (fattr->valid & NFS_ATTR_FATTR_SIZE)
			inode->i_size = nfs_size_to_loff_t(fattr->size);
524
		else
525
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_SIZE);
526
		if (fattr->valid & NFS_ATTR_FATTR_NLINK)
M
Miklos Szeredi 已提交
527
			set_nlink(inode, fattr->nlink);
528
		else if (nfs_server_capable(inode, NFS_CAP_NLINK))
529
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER);
530 531
		if (fattr->valid & NFS_ATTR_FATTR_OWNER)
			inode->i_uid = fattr->uid;
532
		else if (nfs_server_capable(inode, NFS_CAP_OWNER))
533
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER);
534 535
		if (fattr->valid & NFS_ATTR_FATTR_GROUP)
			inode->i_gid = fattr->gid;
536
		else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP))
537
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER);
538 539 540
		if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
			inode->i_blocks = fattr->du.nfs2.blocks;
		if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
L
Linus Torvalds 已提交
541 542 543 544 545
			/*
			 * report the blocks in 512byte units
			 */
			inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
		}
546

547 548 549
		if (nfsi->cache_validity != 0)
			nfsi->cache_validity |= NFS_INO_REVAL_FORCED;

550 551
		nfs_setsecurity(inode, fattr, label);

L
Linus Torvalds 已提交
552
		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
553
		nfsi->attrtimeo_timestamp = now;
554
		nfsi->access_cache = RB_ROOT;
L
Linus Torvalds 已提交
555

556
		nfs_fscache_init_inode(inode);
D
David Howells 已提交
557

L
Linus Torvalds 已提交
558
		unlock_new_inode(inode);
559 560 561 562 563 564 565 566
	} else {
		int err = nfs_refresh_inode(inode, fattr);
		if (err < 0) {
			iput(inode);
			inode = ERR_PTR(err);
			goto out_no_inode;
		}
	}
567
	dprintk("NFS: nfs_fhget(%s/%Lu fh_crc=0x%08x ct=%d)\n",
L
Linus Torvalds 已提交
568
		inode->i_sb->s_id,
569
		(unsigned long long)NFS_FILEID(inode),
570
		nfs_display_fhandle_hash(fh),
L
Linus Torvalds 已提交
571 572 573 574 575 576
		atomic_read(&inode->i_count));

out:
	return inode;

out_no_inode:
577
	dprintk("nfs_fhget: iget failed with error %ld\n", PTR_ERR(inode));
L
Linus Torvalds 已提交
578 579
	goto out;
}
B
Bryan Schumaker 已提交
580
EXPORT_SYMBOL_GPL(nfs_fhget);
L
Linus Torvalds 已提交
581

582
#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET|ATTR_FILE|ATTR_OPEN)
L
Linus Torvalds 已提交
583 584 585 586

int
nfs_setattr(struct dentry *dentry, struct iattr *attr)
{
587
	struct inode *inode = d_inode(dentry);
588
	struct nfs_fattr *fattr;
589
	int error = 0;
L
Linus Torvalds 已提交
590

C
Chuck Lever 已提交
591 592
	nfs_inc_stats(inode, NFSIOS_VFSSETATTR);

593 594 595 596
	/* skip mode change if it's just for clearing setuid/setgid */
	if (attr->ia_valid & (ATTR_KILL_SUID | ATTR_KILL_SGID))
		attr->ia_valid &= ~ATTR_MODE;

L
Linus Torvalds 已提交
597
	if (attr->ia_valid & ATTR_SIZE) {
598 599
		BUG_ON(!S_ISREG(inode->i_mode));

600 601 602 603 604
		error = inode_newsize_ok(inode, attr->ia_size);
		if (error)
			return error;

		if (attr->ia_size == i_size_read(inode))
L
Linus Torvalds 已提交
605 606 607 608 609
			attr->ia_valid &= ~ATTR_SIZE;
	}

	/* Optimization: if the end result is no change, don't RPC */
	attr->ia_valid &= NFS_VALID_ATTRS;
610
	if ((attr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
L
Linus Torvalds 已提交
611 612
		return 0;

613 614
	trace_nfs_setattr_enter(inode);

615
	/* Write all dirty data */
616 617
	if (S_ISREG(inode->i_mode))
		nfs_sync_inode(inode);
618 619

	fattr = nfs_alloc_fattr();
620 621
	if (fattr == NULL) {
		error = -ENOMEM;
622
		goto out;
623 624
	}

625
	error = NFS_PROTO(inode)->setattr(dentry, fattr, attr);
626
	if (error == 0)
627
		error = nfs_refresh_inode(inode, fattr);
628 629
	nfs_free_fattr(fattr);
out:
630
	trace_nfs_setattr_exit(inode, error);
631 632
	return error;
}
B
Bryan Schumaker 已提交
633
EXPORT_SYMBOL_GPL(nfs_setattr);
634

635 636 637 638 639 640 641 642
/**
 * nfs_vmtruncate - unmap mappings "freed" by truncate() syscall
 * @inode: inode of the file used
 * @offset: file offset to start truncating
 *
 * This is a copy of the common vmtruncate, but with the locking
 * corrected to take into account the fact that NFS requires
 * inode->i_size to be updated under the inode->i_lock.
643
 * Note: must be called with inode->i_lock held!
644 645 646
 */
static int nfs_vmtruncate(struct inode * inode, loff_t offset)
{
N
npiggin@suse.de 已提交
647
	int err;
648

N
npiggin@suse.de 已提交
649 650 651
	err = inode_newsize_ok(inode, offset);
	if (err)
		goto out;
652

N
npiggin@suse.de 已提交
653
	i_size_write(inode, offset);
654 655
	/* Optimisation */
	if (offset == 0)
656 657
		NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_DATA |
				NFS_INO_DATA_INVAL_DEFER);
658
	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_SIZE;
N
npiggin@suse.de 已提交
659

660
	spin_unlock(&inode->i_lock);
661
	truncate_pagecache(inode, offset);
662
	spin_lock(&inode->i_lock);
N
npiggin@suse.de 已提交
663 664
out:
	return err;
665 666
}

667 668 669 670
/**
 * nfs_setattr_update_inode - Update inode metadata after a setattr call.
 * @inode: pointer to struct inode
 * @attr: pointer to struct iattr
671
 * @fattr: pointer to struct nfs_fattr
672 673 674 675
 *
 * Note: we do this in the *proc.c in order to ensure that
 *       it works for things like exclusive creates too.
 */
676 677
void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr,
		struct nfs_fattr *fattr)
678
{
679 680 681 682 683
	/* Barrier: bump the attribute generation count. */
	nfs_fattr_set_barrier(fattr);

	spin_lock(&inode->i_lock);
	NFS_I(inode)->attr_gencount = fattr->gencount;
684 685 686 687 688
	if ((attr->ia_valid & ATTR_SIZE) != 0) {
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME);
		nfs_inc_stats(inode, NFSIOS_SETATTRTRUNC);
		nfs_vmtruncate(inode, attr->ia_size);
	}
689
	if ((attr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID)) != 0) {
690
		NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_CTIME;
L
Linus Torvalds 已提交
691
		if ((attr->ia_valid & ATTR_MODE) != 0) {
692 693
			int mode = attr->ia_mode & S_IALLUGO;
			mode |= inode->i_mode & ~S_IALLUGO;
L
Linus Torvalds 已提交
694 695 696 697 698 699
			inode->i_mode = mode;
		}
		if ((attr->ia_valid & ATTR_UID) != 0)
			inode->i_uid = attr->ia_uid;
		if ((attr->ia_valid & ATTR_GID) != 0)
			inode->i_gid = attr->ia_gid;
700
		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
701
			inode->i_ctime = fattr->ctime;
702 703 704
		else
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE
					| NFS_INO_INVALID_CTIME);
705 706
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ACCESS
				| NFS_INO_INVALID_ACL);
707
	}
708 709 710 711
	if (attr->ia_valid & (ATTR_ATIME_SET|ATTR_ATIME)) {
		NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_ATIME
				| NFS_INO_INVALID_CTIME);
		if (fattr->valid & NFS_ATTR_FATTR_ATIME)
712
			inode->i_atime = fattr->atime;
713 714 715 716 717 718
		else if (attr->ia_valid & ATTR_ATIME_SET)
			inode->i_atime = attr->ia_atime;
		else
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME);

		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
719
			inode->i_ctime = fattr->ctime;
720 721 722 723 724 725 726 727
		else
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE
					| NFS_INO_INVALID_CTIME);
	}
	if (attr->ia_valid & (ATTR_MTIME_SET|ATTR_MTIME)) {
		NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_MTIME
				| NFS_INO_INVALID_CTIME);
		if (fattr->valid & NFS_ATTR_FATTR_MTIME)
728
			inode->i_mtime = fattr->mtime;
729 730 731 732 733 734
		else if (attr->ia_valid & ATTR_MTIME_SET)
			inode->i_mtime = attr->ia_mtime;
		else
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME);

		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
735
			inode->i_ctime = fattr->ctime;
736 737 738
		else
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE
					| NFS_INO_INVALID_CTIME);
739
	}
740 741
	if (fattr->valid)
		nfs_update_inode(inode, fattr);
742
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
743
}
B
Bryan Schumaker 已提交
744
EXPORT_SYMBOL_GPL(nfs_setattr_update_inode);
L
Linus Torvalds 已提交
745

746
static void nfs_readdirplus_parent_cache_miss(struct dentry *dentry)
747 748 749
{
	struct dentry *parent;

750 751
	if (!nfs_server_capable(d_inode(dentry), NFS_CAP_READDIRPLUS))
		return;
752
	parent = dget_parent(dentry);
753
	nfs_force_use_readdirplus(d_inode(parent));
754 755 756
	dput(parent);
}

757 758 759 760 761 762 763 764 765 766 767
static void nfs_readdirplus_parent_cache_hit(struct dentry *dentry)
{
	struct dentry *parent;

	if (!nfs_server_capable(d_inode(dentry), NFS_CAP_READDIRPLUS))
		return;
	parent = dget_parent(dentry);
	nfs_advise_use_readdirplus(d_inode(parent));
	dput(parent);
}

768 769 770 771 772 773 774 775 776 777
static bool nfs_need_revalidate_inode(struct inode *inode)
{
	if (NFS_I(inode)->cache_validity &
			(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL))
		return true;
	if (nfs_attribute_cache_expired(inode))
		return true;
	return false;
}

778 779
int nfs_getattr(const struct path *path, struct kstat *stat,
		u32 request_mask, unsigned int query_flags)
L
Linus Torvalds 已提交
780
{
781
	struct inode *inode = d_inode(path->dentry);
782 783
	struct nfs_server *server = NFS_SERVER(inode);
	unsigned long cache_validity;
784
	int err = 0;
785 786
	bool force_sync = query_flags & AT_STATX_FORCE_SYNC;
	bool do_update = false;
L
Linus Torvalds 已提交
787

788
	trace_nfs_getattr_enter(inode);
789 790 791 792

	if ((query_flags & AT_STATX_DONT_SYNC) && !force_sync)
		goto out_no_update;

793
	/* Flush out writes to the server in order to update c/mtime.  */
794 795
	if ((request_mask & (STATX_CTIME|STATX_MTIME)) &&
			S_ISREG(inode->i_mode)) {
796
		err = filemap_write_and_wait(inode->i_mapping);
797 798
		if (err)
			goto out;
799
	}
800 801 802 803 804 805 806

	/*
	 * We may force a getattr if the user cares about atime.
	 *
	 * Note that we only have to check the vfsmount flags here:
	 *  - NFS always sets S_NOATIME by so checking it would give a
	 *    bogus result
807
	 *  - NFS never sets SB_NOATIME or SB_NODIRATIME so there is
808 809
	 *    no point in checking those.
	 */
810 811
	if ((path->mnt->mnt_flags & MNT_NOATIME) ||
	    ((path->mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830
		request_mask &= ~STATX_ATIME;

	/* Is the user requesting attributes that might need revalidation? */
	if (!(request_mask & (STATX_MODE|STATX_NLINK|STATX_ATIME|STATX_CTIME|
					STATX_MTIME|STATX_UID|STATX_GID|
					STATX_SIZE|STATX_BLOCKS)))
		goto out_no_revalidate;

	/* Check whether the cached attributes are stale */
	do_update |= force_sync || nfs_attribute_cache_expired(inode);
	cache_validity = READ_ONCE(NFS_I(inode)->cache_validity);
	do_update |= cache_validity &
		(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL);
	if (request_mask & STATX_ATIME)
		do_update |= cache_validity & NFS_INO_INVALID_ATIME;
	if (request_mask & (STATX_CTIME|STATX_MTIME))
		do_update |= cache_validity & NFS_INO_REVAL_PAGECACHE;
	if (do_update) {
		/* Update the attribute cache */
831 832 833 834
		if (!(server->flags & NFS_MOUNT_NOAC))
			nfs_readdirplus_parent_cache_miss(path->dentry);
		else
			nfs_readdirplus_parent_cache_hit(path->dentry);
835
		err = __nfs_revalidate_inode(server, inode);
836 837
		if (err)
			goto out;
838
	} else
839
		nfs_readdirplus_parent_cache_hit(path->dentry);
840 841 842 843 844 845 846 847
out_no_revalidate:
	/* Only return attributes that were revalidated. */
	stat->result_mask &= request_mask;
out_no_update:
	generic_fillattr(inode, stat);
	stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
	if (S_ISDIR(inode->i_mode))
		stat->blksize = NFS_SERVER(inode)->dtsize;
848
out:
849
	trace_nfs_getattr_exit(inode, err);
L
Linus Torvalds 已提交
850 851
	return err;
}
B
Bryan Schumaker 已提交
852
EXPORT_SYMBOL_GPL(nfs_getattr);
L
Linus Torvalds 已提交
853

854 855
static void nfs_init_lock_context(struct nfs_lock_context *l_ctx)
{
856
	refcount_set(&l_ctx->count, 1);
N
NeilBrown 已提交
857
	l_ctx->lockowner = current->files;
858
	INIT_LIST_HEAD(&l_ctx->list);
859
	atomic_set(&l_ctx->io_count, 0);
860 861 862 863
}

static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx)
{
864
	struct nfs_lock_context *pos;
865

866
	list_for_each_entry_rcu(pos, &ctx->lock_context.list, list) {
N
NeilBrown 已提交
867
		if (pos->lockowner != current->files)
868
			continue;
869 870 871
		if (refcount_inc_not_zero(&pos->count))
			return pos;
	}
872 873 874 875 876 877
	return NULL;
}

struct nfs_lock_context *nfs_get_lock_context(struct nfs_open_context *ctx)
{
	struct nfs_lock_context *res, *new = NULL;
878
	struct inode *inode = d_inode(ctx->dentry);
879

880
	rcu_read_lock();
881
	res = __nfs_find_lock_context(ctx);
882
	rcu_read_unlock();
883 884 885
	if (res == NULL) {
		new = kmalloc(sizeof(*new), GFP_KERNEL);
		if (new == NULL)
886
			return ERR_PTR(-ENOMEM);
887 888 889 890
		nfs_init_lock_context(new);
		spin_lock(&inode->i_lock);
		res = __nfs_find_lock_context(ctx);
		if (res == NULL) {
891 892 893 894 895 896 897 898
			new->open_context = get_nfs_open_context(ctx);
			if (new->open_context) {
				list_add_tail_rcu(&new->list,
						&ctx->lock_context.list);
				res = new;
				new = NULL;
			} else
				res = ERR_PTR(-EBADF);
899
		}
900 901
		spin_unlock(&inode->i_lock);
		kfree(new);
902 903 904
	}
	return res;
}
A
Anna Schumaker 已提交
905
EXPORT_SYMBOL_GPL(nfs_get_lock_context);
906 907 908 909

void nfs_put_lock_context(struct nfs_lock_context *l_ctx)
{
	struct nfs_open_context *ctx = l_ctx->open_context;
910
	struct inode *inode = d_inode(ctx->dentry);
911

912
	if (!refcount_dec_and_lock(&l_ctx->count, &inode->i_lock))
913
		return;
914
	list_del_rcu(&l_ctx->list);
915
	spin_unlock(&inode->i_lock);
916
	put_nfs_open_context(ctx);
917
	kfree_rcu(l_ctx, rcu_head);
918
}
A
Anna Schumaker 已提交
919
EXPORT_SYMBOL_GPL(nfs_put_lock_context);
920

T
Trond Myklebust 已提交
921 922 923 924 925
/**
 * nfs_close_context - Common close_context() routine NFSv2/v3
 * @ctx: pointer to context
 * @is_sync: is this a synchronous close
 *
926 927 928
 * Ensure that the attributes are up to date if we're mounted
 * with close-to-open semantics and we have cached data that will
 * need to be revalidated on open.
T
Trond Myklebust 已提交
929 930 931
 */
void nfs_close_context(struct nfs_open_context *ctx, int is_sync)
{
932
	struct nfs_inode *nfsi;
T
Trond Myklebust 已提交
933 934 935 936 937 938 939
	struct inode *inode;
	struct nfs_server *server;

	if (!(ctx->mode & FMODE_WRITE))
		return;
	if (!is_sync)
		return;
940
	inode = d_inode(ctx->dentry);
941 942
	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
		return;
943 944 945 946 947 948
	nfsi = NFS_I(inode);
	if (inode->i_mapping->nrpages == 0)
		return;
	if (nfsi->cache_validity & NFS_INO_INVALID_DATA)
		return;
	if (!list_empty(&nfsi->open_files))
T
Trond Myklebust 已提交
949 950 951 952 953 954
		return;
	server = NFS_SERVER(inode);
	if (server->flags & NFS_MOUNT_NOCTO)
		return;
	nfs_revalidate_inode(server, inode);
}
B
Bryan Schumaker 已提交
955
EXPORT_SYMBOL_GPL(nfs_close_context);
T
Trond Myklebust 已提交
956

N
NeilBrown 已提交
957 958 959
struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry,
						fmode_t f_mode,
						struct file *filp)
L
Linus Torvalds 已提交
960 961 962
{
	struct nfs_open_context *ctx;

963
	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
964
	if (!ctx)
965 966 967
		return ERR_PTR(-ENOMEM);
	nfs_sb_active(dentry->d_sb);
	ctx->dentry = dget(dentry);
968 969 970 971
	if (filp)
		ctx->cred = get_cred(filp->f_cred);
	else
		ctx->cred = get_current_cred();
972
	ctx->ll_cred = NULL;
973 974 975 976
	ctx->state = NULL;
	ctx->mode = f_mode;
	ctx->flags = 0;
	ctx->error = 0;
N
NeilBrown 已提交
977
	ctx->flock_owner = (fl_owner_t)filp;
978 979 980
	nfs_init_lock_context(&ctx->lock_context);
	ctx->lock_context.open_context = ctx;
	INIT_LIST_HEAD(&ctx->list);
981
	ctx->mdsthreshold = NULL;
L
Linus Torvalds 已提交
982 983
	return ctx;
}
B
Bryan Schumaker 已提交
984
EXPORT_SYMBOL_GPL(alloc_nfs_open_context);
L
Linus Torvalds 已提交
985 986 987

struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx)
{
988 989 990
	if (ctx != NULL && refcount_inc_not_zero(&ctx->lock_context.count))
		return ctx;
	return NULL;
L
Linus Torvalds 已提交
991
}
B
Bryan Schumaker 已提交
992
EXPORT_SYMBOL_GPL(get_nfs_open_context);
L
Linus Torvalds 已提交
993

T
Trond Myklebust 已提交
994
static void __put_nfs_open_context(struct nfs_open_context *ctx, int is_sync)
L
Linus Torvalds 已提交
995
{
996
	struct inode *inode = d_inode(ctx->dentry);
997
	struct super_block *sb = ctx->dentry->d_sb;
998

999 1000
	if (!refcount_dec_and_test(&ctx->lock_context.count))
		return;
1001
	if (!list_empty(&ctx->list)) {
1002 1003
		spin_lock(&inode->i_lock);
		list_del_rcu(&ctx->list);
1004
		spin_unlock(&inode->i_lock);
1005
	}
1006 1007
	if (inode != NULL)
		NFS_PROTO(inode)->close_context(ctx, is_sync);
1008
	put_cred(ctx->cred);
1009
	dput(ctx->dentry);
1010
	nfs_sb_deactive(sb);
1011
	put_rpccred(ctx->ll_cred);
1012
	kfree(ctx->mdsthreshold);
1013
	kfree_rcu(ctx, rcu_head);
1014 1015
}

1016 1017 1018 1019
void put_nfs_open_context(struct nfs_open_context *ctx)
{
	__put_nfs_open_context(ctx, 0);
}
B
Bryan Schumaker 已提交
1020
EXPORT_SYMBOL_GPL(put_nfs_open_context);
1021

1022 1023 1024 1025 1026
static void put_nfs_open_context_sync(struct nfs_open_context *ctx)
{
	__put_nfs_open_context(ctx, 1);
}

L
Linus Torvalds 已提交
1027 1028 1029 1030
/*
 * Ensure that mmap has a recent RPC credential for use when writing out
 * shared pages
 */
1031
void nfs_inode_attach_open_context(struct nfs_open_context *ctx)
L
Linus Torvalds 已提交
1032
{
1033
	struct inode *inode = d_inode(ctx->dentry);
L
Linus Torvalds 已提交
1034 1035 1036
	struct nfs_inode *nfsi = NFS_I(inode);

	spin_lock(&inode->i_lock);
1037 1038 1039 1040
	if (list_empty(&nfsi->open_files) &&
	    (nfsi->cache_validity & NFS_INO_DATA_INVAL_DEFER))
		nfsi->cache_validity |= NFS_INO_INVALID_DATA |
			NFS_INO_REVAL_FORCED;
1041
	list_add_tail_rcu(&ctx->list, &nfsi->open_files);
L
Linus Torvalds 已提交
1042 1043
	spin_unlock(&inode->i_lock);
}
1044 1045 1046 1047 1048 1049 1050 1051
EXPORT_SYMBOL_GPL(nfs_inode_attach_open_context);

void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx)
{
	filp->private_data = get_nfs_open_context(ctx);
	if (list_empty(&ctx->list))
		nfs_inode_attach_open_context(ctx);
}
B
Bryan Schumaker 已提交
1052
EXPORT_SYMBOL_GPL(nfs_file_set_open_context);
L
Linus Torvalds 已提交
1053

1054 1055 1056
/*
 * Given an inode, search for an open context with the desired characteristics
 */
1057
struct nfs_open_context *nfs_find_open_context(struct inode *inode, const struct cred *cred, fmode_t mode)
L
Linus Torvalds 已提交
1058 1059 1060 1061
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_open_context *pos, *ctx = NULL;

1062 1063
	rcu_read_lock();
	list_for_each_entry_rcu(pos, &nfsi->open_files, list) {
1064
		if (cred != NULL && cred_fscmp(pos->cred, cred) != 0)
1065
			continue;
1066 1067 1068
		if ((pos->mode & (FMODE_READ|FMODE_WRITE)) != mode)
			continue;
		ctx = get_nfs_open_context(pos);
1069 1070
		if (ctx)
			break;
L
Linus Torvalds 已提交
1071
	}
1072
	rcu_read_unlock();
L
Linus Torvalds 已提交
1073 1074 1075
	return ctx;
}

A
Anna Schumaker 已提交
1076
void nfs_file_clear_open_context(struct file *filp)
L
Linus Torvalds 已提交
1077
{
1078
	struct nfs_open_context *ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
1079 1080

	if (ctx) {
1081
		struct inode *inode = d_inode(ctx->dentry);
1082

1083 1084 1085 1086 1087 1088
		/*
		 * We fatal error on write before. Try to writeback
		 * every page again.
		 */
		if (ctx->error < 0)
			invalidate_inode_pages2(inode->i_mapping);
L
Linus Torvalds 已提交
1089
		filp->private_data = NULL;
1090
		put_nfs_open_context_sync(ctx);
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
	}
}

/*
 * These allocate and release file read/write context information.
 */
int nfs_open(struct inode *inode, struct file *filp)
{
	struct nfs_open_context *ctx;

N
NeilBrown 已提交
1101
	ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode, filp);
1102 1103
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
L
Linus Torvalds 已提交
1104 1105
	nfs_file_set_open_context(filp, ctx);
	put_nfs_open_context(ctx);
1106
	nfs_fscache_open_file(inode, filp);
L
Linus Torvalds 已提交
1107 1108
	return 0;
}
1109
EXPORT_SYMBOL_GPL(nfs_open);
L
Linus Torvalds 已提交
1110 1111 1112 1113 1114 1115 1116 1117 1118

/*
 * This function is called whenever some part of NFS notices that
 * the cached attributes have to be refreshed.
 */
int
__nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
{
	int		 status = -ESTALE;
1119
	struct nfs4_label *label = NULL;
1120
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
1121 1122
	struct nfs_inode *nfsi = NFS_I(inode);

1123 1124
	dfprintk(PAGECACHE, "NFS: revalidating (%s/%Lu)\n",
		inode->i_sb->s_id, (unsigned long long)NFS_FILEID(inode));
L
Linus Torvalds 已提交
1125

1126 1127
	trace_nfs_revalidate_inode_enter(inode);

1128
	if (is_bad_inode(inode))
1129
		goto out;
L
Linus Torvalds 已提交
1130
	if (NFS_STALE(inode))
1131
		goto out;
1132

1133 1134 1135 1136 1137 1138 1139
	/* pNFS: Attributes aren't updated until we layoutcommit */
	if (S_ISREG(inode->i_mode)) {
		status = pnfs_sync_inode(inode, false);
		if (status)
			goto out;
	}

1140 1141 1142 1143 1144
	status = -ENOMEM;
	fattr = nfs_alloc_fattr();
	if (fattr == NULL)
		goto out;

1145
	nfs_inc_stats(inode, NFSIOS_INODEREVALIDATE);
1146 1147 1148 1149 1150 1151 1152

	label = nfs4_label_alloc(NFS_SERVER(inode), GFP_KERNEL);
	if (IS_ERR(label)) {
		status = PTR_ERR(label);
		goto out;
	}

1153 1154
	status = NFS_PROTO(inode)->getattr(server, NFS_FH(inode), fattr,
			label, inode);
L
Linus Torvalds 已提交
1155
	if (status != 0) {
1156
		dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) getattr failed, error=%d\n",
L
Linus Torvalds 已提交
1157
			 inode->i_sb->s_id,
1158
			 (unsigned long long)NFS_FILEID(inode), status);
1159 1160 1161 1162 1163 1164 1165
		switch (status) {
		case -ETIMEDOUT:
			/* A soft timeout occurred. Use cached information? */
			if (server->flags & NFS_MOUNT_SOFTREVAL)
				status = 0;
			break;
		case -ESTALE:
L
Linus Torvalds 已提交
1166 1167
			nfs_zap_caches(inode);
			if (!S_ISDIR(inode->i_mode))
B
Benny Halevy 已提交
1168
				set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
1169
		}
1170
		goto err_out;
L
Linus Torvalds 已提交
1171 1172
	}

1173
	status = nfs_refresh_inode(inode, fattr);
L
Linus Torvalds 已提交
1174
	if (status) {
1175
		dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) refresh failed, error=%d\n",
L
Linus Torvalds 已提交
1176
			 inode->i_sb->s_id,
1177
			 (unsigned long long)NFS_FILEID(inode), status);
1178
		goto err_out;
L
Linus Torvalds 已提交
1179
	}
1180

1181
	if (nfsi->cache_validity & NFS_INO_INVALID_ACL)
1182
		nfs_zap_acl_cache(inode);
1183

1184 1185
	nfs_setsecurity(inode, fattr, label);

1186
	dfprintk(PAGECACHE, "NFS: (%s/%Lu) revalidation complete\n",
L
Linus Torvalds 已提交
1187
		inode->i_sb->s_id,
1188
		(unsigned long long)NFS_FILEID(inode));
L
Linus Torvalds 已提交
1189

1190 1191 1192
err_out:
	nfs4_label_free(label);
out:
1193
	nfs_free_fattr(fattr);
1194
	trace_nfs_revalidate_inode_exit(inode, status);
L
Linus Torvalds 已提交
1195 1196 1197
	return status;
}

1198
int nfs_attribute_cache_expired(struct inode *inode)
1199
{
1200
	if (nfs_have_delegated_attributes(inode))
L
Linus Torvalds 已提交
1201
		return 0;
1202
	return nfs_attribute_timeout(inode);
L
Linus Torvalds 已提交
1203 1204 1205 1206
}

/**
 * nfs_revalidate_inode - Revalidate the inode attributes
1207 1208
 * @server: pointer to nfs_server struct
 * @inode: pointer to inode struct
L
Linus Torvalds 已提交
1209 1210 1211 1212 1213
 *
 * Updates inode attribute information by retrieving the data from the server.
 */
int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
{
1214
	if (!nfs_need_revalidate_inode(inode))
L
Linus Torvalds 已提交
1215 1216 1217
		return NFS_STALE(inode) ? -ESTALE : 0;
	return __nfs_revalidate_inode(server, inode);
}
B
Bryan Schumaker 已提交
1218
EXPORT_SYMBOL_GPL(nfs_revalidate_inode);
L
Linus Torvalds 已提交
1219

1220
static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping)
1221 1222
{
	struct nfs_inode *nfsi = NFS_I(inode);
1223 1224
	int ret;

1225
	if (mapping->nrpages != 0) {
1226 1227 1228 1229 1230 1231
		if (S_ISREG(inode->i_mode)) {
			ret = nfs_sync_mapping(mapping);
			if (ret < 0)
				return ret;
		}
		ret = invalidate_inode_pages2(mapping);
1232 1233 1234
		if (ret < 0)
			return ret;
	}
1235 1236
	if (S_ISDIR(inode->i_mode)) {
		spin_lock(&inode->i_lock);
1237
		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
1238 1239
		spin_unlock(&inode->i_lock);
	}
1240
	nfs_inc_stats(inode, NFSIOS_DATAINVALIDATE);
D
David Howells 已提交
1241
	nfs_fscache_wait_on_invalidate(inode);
1242

1243 1244 1245
	dfprintk(PAGECACHE, "NFS: (%s/%Lu) data cache invalidated\n",
			inode->i_sb->s_id,
			(unsigned long long)NFS_FILEID(inode));
1246 1247 1248
	return 0;
}

1249
bool nfs_mapping_need_revalidate_inode(struct inode *inode)
1250
{
1251 1252
	return nfs_check_cache_invalid(inode, NFS_INO_REVAL_PAGECACHE) ||
		NFS_STALE(inode);
1253 1254
}

A
Al Viro 已提交
1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
int nfs_revalidate_mapping_rcu(struct inode *inode)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	unsigned long *bitlock = &nfsi->flags;
	int ret = 0;

	if (IS_SWAPFILE(inode))
		goto out;
	if (nfs_mapping_need_revalidate_inode(inode)) {
		ret = -ECHILD;
		goto out;
	}
	spin_lock(&inode->i_lock);
	if (test_bit(NFS_INO_INVALIDATING, bitlock) ||
	    (nfsi->cache_validity & NFS_INO_INVALID_DATA))
		ret = -ECHILD;
	spin_unlock(&inode->i_lock);
out:
	return ret;
}

1276
/**
1277
 * nfs_revalidate_mapping - Revalidate the pagecache
1278 1279
 * @inode: pointer to host inode
 * @mapping: pointer to mapping
1280
 */
1281 1282
int nfs_revalidate_mapping(struct inode *inode,
		struct address_space *mapping)
1283 1284
{
	struct nfs_inode *nfsi = NFS_I(inode);
1285
	unsigned long *bitlock = &nfsi->flags;
1286
	int ret = 0;
1287

1288 1289 1290 1291
	/* swapfiles are not supposed to be shared. */
	if (IS_SWAPFILE(inode))
		goto out;

1292
	if (nfs_mapping_need_revalidate_inode(inode)) {
1293 1294 1295
		ret = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
		if (ret < 0)
			goto out;
1296
	}
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310

	/*
	 * We must clear NFS_INO_INVALID_DATA first to ensure that
	 * invalidations that come in while we're shooting down the mappings
	 * are respected. But, that leaves a race window where one revalidator
	 * can clear the flag, and then another checks it before the mapping
	 * gets invalidated. Fix that by serializing access to this part of
	 * the function.
	 *
	 * At the same time, we need to allow other tasks to see whether we
	 * might be in the middle of invalidating the pages, so we only set
	 * the bit lock here if it looks like we're going to be doing that.
	 */
	for (;;) {
1311 1312
		ret = wait_on_bit_action(bitlock, NFS_INO_INVALIDATING,
					 nfs_wait_bit_killable, TASK_KILLABLE);
1313 1314
		if (ret)
			goto out;
1315 1316 1317 1318 1319 1320
		spin_lock(&inode->i_lock);
		if (test_bit(NFS_INO_INVALIDATING, bitlock)) {
			spin_unlock(&inode->i_lock);
			continue;
		}
		if (nfsi->cache_validity & NFS_INO_INVALID_DATA)
1321 1322
			break;
		spin_unlock(&inode->i_lock);
1323
		goto out;
1324 1325
	}

1326
	set_bit(NFS_INO_INVALIDATING, bitlock);
1327
	smp_wmb();
1328 1329
	nfsi->cache_validity &= ~(NFS_INO_INVALID_DATA|
			NFS_INO_DATA_INVAL_DEFER);
1330 1331
	spin_unlock(&inode->i_lock);
	trace_nfs_invalidate_mapping_enter(inode);
1332
	ret = nfs_invalidate_mapping(inode, mapping);
1333 1334
	trace_nfs_invalidate_mapping_exit(inode, ret);

1335
	clear_bit_unlock(NFS_INO_INVALIDATING, bitlock);
1336
	smp_mb__after_atomic();
1337
	wake_up_bit(bitlock, NFS_INO_INVALIDATING);
1338
out:
1339
	return ret;
1340 1341
}

1342
static bool nfs_file_has_writers(struct nfs_inode *nfsi)
1343
{
1344 1345 1346 1347 1348 1349
	struct inode *inode = &nfsi->vfs_inode;

	if (!S_ISREG(inode->i_mode))
		return false;
	if (list_empty(&nfsi->open_files))
		return false;
1350
	return inode_is_open_for_write(inode);
1351 1352
}

1353
static bool nfs_file_has_buffered_writers(struct nfs_inode *nfsi)
1354
{
1355
	return nfs_file_has_writers(nfsi) && nfs_file_io_is_buffered(nfsi);
1356 1357
}

1358
static void nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1359
{
1360
	struct timespec64 ts;
1361

1362 1363
	if ((fattr->valid & NFS_ATTR_FATTR_PRECHANGE)
			&& (fattr->valid & NFS_ATTR_FATTR_CHANGE)
1364
			&& inode_eq_iversion_raw(inode, fattr->pre_change_attr)) {
J
Jeff Layton 已提交
1365
		inode_set_iversion_raw(inode, fattr->change_attr);
1366
		if (S_ISDIR(inode->i_mode))
1367
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
1368
	}
1369
	/* If we have atomic WCC data, we may update some attributes */
1370
	ts = inode->i_ctime;
1371 1372
	if ((fattr->valid & NFS_ATTR_FATTR_PRECTIME)
			&& (fattr->valid & NFS_ATTR_FATTR_CTIME)
1373 1374
			&& timespec64_equal(&ts, &fattr->pre_ctime)) {
		inode->i_ctime = fattr->ctime;
1375
	}
1376

1377
	ts = inode->i_mtime;
1378 1379
	if ((fattr->valid & NFS_ATTR_FATTR_PREMTIME)
			&& (fattr->valid & NFS_ATTR_FATTR_MTIME)
1380 1381
			&& timespec64_equal(&ts, &fattr->pre_mtime)) {
		inode->i_mtime = fattr->mtime;
1382
		if (S_ISDIR(inode->i_mode))
1383
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
1384
	}
1385 1386 1387
	if ((fattr->valid & NFS_ATTR_FATTR_PRESIZE)
			&& (fattr->valid & NFS_ATTR_FATTR_SIZE)
			&& i_size_read(inode) == nfs_size_to_loff_t(fattr->pre_size)
1388
			&& !nfs_have_writebacks(inode)) {
1389 1390
		i_size_write(inode, nfs_size_to_loff_t(fattr->size));
	}
1391 1392
}

L
Linus Torvalds 已提交
1393
/**
1394
 * nfs_check_inode_attributes - verify consistency of the inode attribute cache
1395 1396
 * @inode: pointer to inode
 * @fattr: updated attributes
L
Linus Torvalds 已提交
1397 1398 1399 1400 1401
 *
 * Verifies the attribute cache. If we have just changed the attributes,
 * so that fattr carries weak cache consistency data, then it may
 * also update the ctime/mtime/change_attribute.
 */
1402
static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fattr)
L
Linus Torvalds 已提交
1403 1404 1405
{
	struct nfs_inode *nfsi = NFS_I(inode);
	loff_t cur_size, new_isize;
1406
	unsigned long invalid = 0;
1407
	struct timespec64 ts;
1408

1409
	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
1410
		return 0;
1411

1412 1413 1414 1415
	if (!(fattr->valid & NFS_ATTR_FATTR_FILEID)) {
		/* Only a mounted-on-fileid? Just exit */
		if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID)
			return 0;
1416
	/* Has the inode gone and changed behind our back? */
1417
	} else if (nfsi->fileid != fattr->fileid) {
1418 1419 1420 1421
		/* Is this perhaps the mounted-on fileid? */
		if ((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) &&
		    nfsi->fileid == fattr->mounted_on_fileid)
			return 0;
1422
		return -ESTALE;
1423
	}
1424
	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
1425
		return -ESTALE;
1426

1427

1428
	if (!nfs_file_has_buffered_writers(nfsi)) {
1429
		/* Verify a few of the more important attributes */
1430
		if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 && !inode_eq_iversion_raw(inode, fattr->change_attr))
1431 1432
			invalid |= NFS_INO_INVALID_CHANGE
				| NFS_INO_REVAL_PAGECACHE;
L
Linus Torvalds 已提交
1433

1434 1435
		ts = inode->i_mtime;
		if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec64_equal(&ts, &fattr->mtime))
1436
			invalid |= NFS_INO_INVALID_MTIME;
1437

1438 1439
		ts = inode->i_ctime;
		if ((fattr->valid & NFS_ATTR_FATTR_CTIME) && !timespec64_equal(&ts, &fattr->ctime))
1440
			invalid |= NFS_INO_INVALID_CTIME;
1441 1442 1443 1444 1445

		if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
			cur_size = i_size_read(inode);
			new_isize = nfs_size_to_loff_t(fattr->size);
			if (cur_size != new_isize)
1446 1447
				invalid |= NFS_INO_INVALID_SIZE
					| NFS_INO_REVAL_PAGECACHE;
1448
		}
1449
	}
L
Linus Torvalds 已提交
1450 1451

	/* Have any file permissions changed? */
1452
	if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
1453 1454 1455
		invalid |= NFS_INO_INVALID_ACCESS
			| NFS_INO_INVALID_ACL
			| NFS_INO_INVALID_OTHER;
1456
	if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && !uid_eq(inode->i_uid, fattr->uid))
1457 1458 1459
		invalid |= NFS_INO_INVALID_ACCESS
			| NFS_INO_INVALID_ACL
			| NFS_INO_INVALID_OTHER;
1460
	if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && !gid_eq(inode->i_gid, fattr->gid))
1461 1462 1463
		invalid |= NFS_INO_INVALID_ACCESS
			| NFS_INO_INVALID_ACL
			| NFS_INO_INVALID_OTHER;
L
Linus Torvalds 已提交
1464 1465

	/* Has the link count changed? */
1466
	if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink)
1467
		invalid |= NFS_INO_INVALID_OTHER;
L
Linus Torvalds 已提交
1468

1469 1470
	ts = inode->i_atime;
	if ((fattr->valid & NFS_ATTR_FATTR_ATIME) && !timespec64_equal(&ts, &fattr->atime))
1471 1472 1473
		invalid |= NFS_INO_INVALID_ATIME;

	if (invalid != 0)
1474
		nfs_set_cache_invalid(inode, invalid);
L
Linus Torvalds 已提交
1475

1476
	nfsi->read_cache_jiffies = fattr->time_start;
L
Linus Torvalds 已提交
1477 1478 1479
	return 0;
}

1480
static atomic_long_t nfs_attr_generation_counter;
1481 1482 1483

static unsigned long nfs_read_attr_generation_counter(void)
{
1484
	return atomic_long_read(&nfs_attr_generation_counter);
1485 1486 1487 1488
}

unsigned long nfs_inc_attr_generation_counter(void)
{
1489
	return atomic_long_inc_return(&nfs_attr_generation_counter);
1490
}
1491
EXPORT_SYMBOL_GPL(nfs_inc_attr_generation_counter);
1492 1493 1494 1495 1496 1497

void nfs_fattr_init(struct nfs_fattr *fattr)
{
	fattr->valid = 0;
	fattr->time_start = jiffies;
	fattr->gencount = nfs_inc_attr_generation_counter();
1498 1499
	fattr->owner_name = NULL;
	fattr->group_name = NULL;
1500
}
B
Bryan Schumaker 已提交
1501
EXPORT_SYMBOL_GPL(nfs_fattr_init);
1502

1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
/**
 * nfs_fattr_set_barrier
 * @fattr: attributes
 *
 * Used to set a barrier after an attribute was updated. This
 * barrier ensures that older attributes from RPC calls that may
 * have raced with our update cannot clobber these new values.
 * Note that you are still responsible for ensuring that other
 * operations which change the attribute on the server do not
 * collide.
 */
void nfs_fattr_set_barrier(struct nfs_fattr *fattr)
{
	fattr->gencount = nfs_inc_attr_generation_counter();
}

1519 1520 1521 1522 1523 1524 1525 1526 1527
struct nfs_fattr *nfs_alloc_fattr(void)
{
	struct nfs_fattr *fattr;

	fattr = kmalloc(sizeof(*fattr), GFP_NOFS);
	if (fattr != NULL)
		nfs_fattr_init(fattr);
	return fattr;
}
B
Bryan Schumaker 已提交
1528
EXPORT_SYMBOL_GPL(nfs_alloc_fattr);
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538

struct nfs_fh *nfs_alloc_fhandle(void)
{
	struct nfs_fh *fh;

	fh = kmalloc(sizeof(struct nfs_fh), GFP_NOFS);
	if (fh != NULL)
		fh->size = 0;
	return fh;
}
B
Bryan Schumaker 已提交
1539
EXPORT_SYMBOL_GPL(nfs_alloc_fhandle);
1540

1541
#ifdef NFS_DEBUG
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
/*
 * _nfs_display_fhandle_hash - calculate the crc32 hash for the filehandle
 *                             in the same way that wireshark does
 *
 * @fh: file handle
 *
 * For debugging only.
 */
u32 _nfs_display_fhandle_hash(const struct nfs_fh *fh)
{
	/* wireshark uses 32-bit AUTODIN crc and does a bitwise
	 * not on the result */
1554
	return nfs_fhandle_hash(fh);
1555
}
C
Chuck Lever 已提交
1556
EXPORT_SYMBOL_GPL(_nfs_display_fhandle_hash);
1557 1558

/*
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
 * _nfs_display_fhandle - display an NFS file handle on the console
 *
 * @fh: file handle to display
 * @caption: display caption
 *
 * For debugging only.
 */
void _nfs_display_fhandle(const struct nfs_fh *fh, const char *caption)
{
	unsigned short i;

1570
	if (fh == NULL || fh->size == 0) {
1571 1572 1573 1574
		printk(KERN_DEFAULT "%s at %p is empty\n", caption, fh);
		return;
	}

1575 1576
	printk(KERN_DEFAULT "%s at %p is %u bytes, crc: 0x%08x:\n",
	       caption, fh, fh->size, _nfs_display_fhandle_hash(fh));
1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	for (i = 0; i < fh->size; i += 16) {
		__be32 *pos = (__be32 *)&fh->data[i];

		switch ((fh->size - i - 1) >> 2) {
		case 0:
			printk(KERN_DEFAULT " %08x\n",
				be32_to_cpup(pos));
			break;
		case 1:
			printk(KERN_DEFAULT " %08x %08x\n",
				be32_to_cpup(pos), be32_to_cpup(pos + 1));
			break;
		case 2:
			printk(KERN_DEFAULT " %08x %08x %08x\n",
				be32_to_cpup(pos), be32_to_cpup(pos + 1),
				be32_to_cpup(pos + 2));
			break;
		default:
			printk(KERN_DEFAULT " %08x %08x %08x %08x\n",
				be32_to_cpup(pos), be32_to_cpup(pos + 1),
				be32_to_cpup(pos + 2), be32_to_cpup(pos + 3));
		}
	}
}
C
Chuck Lever 已提交
1601
EXPORT_SYMBOL_GPL(_nfs_display_fhandle);
1602 1603
#endif

1604 1605
/**
 * nfs_inode_attrs_need_update - check if the inode attributes need updating
1606 1607
 * @inode: pointer to inode
 * @fattr: attributes
1608 1609 1610 1611 1612 1613 1614 1615 1616
 *
 * Attempt to divine whether or not an RPC call reply carrying stale
 * attributes got scheduled after another call carrying updated ones.
 *
 * To do so, the function first assumes that a more recent ctime means
 * that the attributes in fattr are newer, however it also attempt to
 * catch the case where ctime either didn't change, or went backwards
 * (if someone reset the clock on the server) by looking at whether
 * or not this RPC call was started after the inode was last updated.
1617
 * Note also the check for wraparound of 'attr_gencount'
1618 1619 1620 1621 1622 1623 1624 1625 1626
 *
 * The function returns 'true' if it thinks the attributes in 'fattr' are
 * more recent than the ones cached in the inode.
 *
 */
static int nfs_inode_attrs_need_update(const struct inode *inode, const struct nfs_fattr *fattr)
{
	const struct nfs_inode *nfsi = NFS_I(inode);

1627 1628
	return ((long)fattr->gencount - (long)nfsi->attr_gencount) > 0 ||
		((long)nfsi->attr_gencount - (long)nfs_read_attr_generation_counter() > 0);
1629 1630 1631 1632
}

static int nfs_refresh_inode_locked(struct inode *inode, struct nfs_fattr *fattr)
{
1633 1634 1635 1636
	int ret;

	trace_nfs_refresh_inode_enter(inode);

1637
	if (nfs_inode_attrs_need_update(inode, fattr))
1638 1639 1640 1641 1642 1643
		ret = nfs_update_inode(inode, fattr);
	else
		ret = nfs_check_inode_attributes(inode, fattr);

	trace_nfs_refresh_inode_exit(inode, ret);
	return ret;
1644 1645
}

1646 1647
/**
 * nfs_refresh_inode - try to update the inode attribute cache
1648 1649
 * @inode: pointer to inode
 * @fattr: updated attributes
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662
 *
 * Check that an RPC call that returned attributes has not overlapped with
 * other recent updates of the inode metadata, then decide whether it is
 * safe to do a full update of the inode attributes, or whether just to
 * call nfs_check_inode_attributes.
 */
int nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
{
	int status;

	if ((fattr->valid & NFS_ATTR_FATTR) == 0)
		return 0;
	spin_lock(&inode->i_lock);
1663
	status = nfs_refresh_inode_locked(inode, fattr);
1664
	spin_unlock(&inode->i_lock);
D
David Howells 已提交
1665

1666 1667
	return status;
}
B
Bryan Schumaker 已提交
1668
EXPORT_SYMBOL_GPL(nfs_refresh_inode);
1669

1670 1671
static int nfs_post_op_update_inode_locked(struct inode *inode,
		struct nfs_fattr *fattr, unsigned int invalid)
1672
{
1673 1674 1675
	if (S_ISDIR(inode->i_mode))
		invalid |= NFS_INO_INVALID_DATA;
	nfs_set_cache_invalid(inode, invalid);
1676 1677 1678 1679 1680
	if ((fattr->valid & NFS_ATTR_FATTR) == 0)
		return 0;
	return nfs_refresh_inode_locked(inode, fattr);
}

1681 1682
/**
 * nfs_post_op_update_inode - try to update the inode attribute cache
1683 1684
 * @inode: pointer to inode
 * @fattr: updated attributes
1685 1686 1687
 *
 * After an operation that has changed the inode metadata, mark the
 * attribute cache as being invalid, then try to update it.
1688 1689 1690 1691 1692 1693
 *
 * NB: if the server didn't return any post op attributes, this
 * function will force the retrieval of attributes before the next
 * NFS request.  Thus it should be used only for operations that
 * are expected to change one or more attributes, to avoid
 * unnecessary NFS requests and trips through nfs_update_inode().
1694 1695 1696
 */
int nfs_post_op_update_inode(struct inode *inode, struct nfs_fattr *fattr)
{
1697
	int status;
1698

1699
	spin_lock(&inode->i_lock);
1700
	nfs_fattr_set_barrier(fattr);
1701 1702
	status = nfs_post_op_update_inode_locked(inode, fattr,
			NFS_INO_INVALID_CHANGE
1703 1704
			| NFS_INO_INVALID_CTIME
			| NFS_INO_REVAL_FORCED);
1705
	spin_unlock(&inode->i_lock);
1706

1707
	return status;
1708
}
B
Bryan Schumaker 已提交
1709
EXPORT_SYMBOL_GPL(nfs_post_op_update_inode);
1710

1711
/**
1712
 * nfs_post_op_update_inode_force_wcc_locked - update the inode attribute cache
1713 1714
 * @inode: pointer to inode
 * @fattr: updated attributes
1715 1716 1717 1718 1719 1720 1721
 *
 * After an operation that has changed the inode metadata, mark the
 * attribute cache as being invalid, then try to update it. Fake up
 * weak cache consistency data, if none exist.
 *
 * This function is mainly designed to be used by the ->write_done() functions.
 */
1722
int nfs_post_op_update_inode_force_wcc_locked(struct inode *inode, struct nfs_fattr *fattr)
1723
{
1724 1725 1726 1727 1728
	int status;

	/* Don't do a WCC update if these attributes are already stale */
	if ((fattr->valid & NFS_ATTR_FATTR) == 0 ||
			!nfs_inode_attrs_need_update(inode, fattr)) {
1729 1730 1731 1732
		fattr->valid &= ~(NFS_ATTR_FATTR_PRECHANGE
				| NFS_ATTR_FATTR_PRESIZE
				| NFS_ATTR_FATTR_PREMTIME
				| NFS_ATTR_FATTR_PRECTIME);
1733 1734
		goto out_noforce;
	}
1735 1736
	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRECHANGE) == 0) {
J
Jeff Layton 已提交
1737
		fattr->pre_change_attr = inode_peek_iversion_raw(inode);
1738
		fattr->valid |= NFS_ATTR_FATTR_PRECHANGE;
1739
	}
1740 1741
	if ((fattr->valid & NFS_ATTR_FATTR_CTIME) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRECTIME) == 0) {
1742
		fattr->pre_ctime = inode->i_ctime;
1743 1744 1745 1746
		fattr->valid |= NFS_ATTR_FATTR_PRECTIME;
	}
	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PREMTIME) == 0) {
1747
		fattr->pre_mtime = inode->i_mtime;
1748 1749 1750 1751
		fattr->valid |= NFS_ATTR_FATTR_PREMTIME;
	}
	if ((fattr->valid & NFS_ATTR_FATTR_SIZE) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRESIZE) == 0) {
1752
		fattr->pre_size = i_size_read(inode);
1753
		fattr->valid |= NFS_ATTR_FATTR_PRESIZE;
1754
	}
1755
out_noforce:
1756 1757 1758 1759
	status = nfs_post_op_update_inode_locked(inode, fattr,
			NFS_INO_INVALID_CHANGE
			| NFS_INO_INVALID_CTIME
			| NFS_INO_INVALID_MTIME);
1760 1761 1762 1763 1764
	return status;
}

/**
 * nfs_post_op_update_inode_force_wcc - try to update the inode attribute cache
1765 1766
 * @inode: pointer to inode
 * @fattr: updated attributes
1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778
 *
 * After an operation that has changed the inode metadata, mark the
 * attribute cache as being invalid, then try to update it. Fake up
 * weak cache consistency data, if none exist.
 *
 * This function is mainly designed to be used by the ->write_done() functions.
 */
int nfs_post_op_update_inode_force_wcc(struct inode *inode, struct nfs_fattr *fattr)
{
	int status;

	spin_lock(&inode->i_lock);
1779
	nfs_fattr_set_barrier(fattr);
1780
	status = nfs_post_op_update_inode_force_wcc_locked(inode, fattr);
1781 1782
	spin_unlock(&inode->i_lock);
	return status;
1783
}
B
Bryan Schumaker 已提交
1784
EXPORT_SYMBOL_GPL(nfs_post_op_update_inode_force_wcc);
1785

1786

L
Linus Torvalds 已提交
1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
/*
 * Many nfs protocol calls return the new file attributes after
 * an operation.  Here we update the inode to reflect the state
 * of the server's inode.
 *
 * This is a bit tricky because we have to make sure all dirty pages
 * have been sent off to the server before calling invalidate_inode_pages.
 * To make sure no other process adds more write requests while we try
 * our best to flush them, we make them sleep during the attribute refresh.
 *
 * A very similar scenario holds for the dir cache.
 */
1799
static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
L
Linus Torvalds 已提交
1800
{
1801
	struct nfs_server *server;
L
Linus Torvalds 已提交
1802
	struct nfs_inode *nfsi = NFS_I(inode);
1803
	loff_t cur_isize, new_isize;
1804
	unsigned long invalid = 0;
1805
	unsigned long now = jiffies;
1806
	unsigned long save_cache_validity;
1807
	bool have_writers = nfs_file_has_buffered_writers(nfsi);
1808
	bool cache_revalidated = true;
1809
	bool attr_changed = false;
1810
	bool have_delegation;
L
Linus Torvalds 已提交
1811

1812
	dfprintk(VFS, "NFS: %s(%s/%lu fh_crc=0x%08x ct=%d info=0x%x)\n",
1813
			__func__, inode->i_sb->s_id, inode->i_ino,
1814
			nfs_display_fhandle_hash(NFS_FH(inode)),
L
Linus Torvalds 已提交
1815 1816
			atomic_read(&inode->i_count), fattr->valid);

1817 1818 1819 1820
	if (!(fattr->valid & NFS_ATTR_FATTR_FILEID)) {
		/* Only a mounted-on-fileid? Just exit */
		if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID)
			return 0;
1821
	/* Has the inode gone and changed behind our back? */
1822
	} else if (nfsi->fileid != fattr->fileid) {
1823 1824 1825 1826
		/* Is this perhaps the mounted-on fileid? */
		if ((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) &&
		    nfsi->fileid == fattr->mounted_on_fileid)
			return 0;
1827 1828 1829 1830 1831 1832 1833
		printk(KERN_ERR "NFS: server %s error: fileid changed\n"
			"fsid %s: expected fileid 0x%Lx, got 0x%Lx\n",
			NFS_SERVER(inode)->nfs_client->cl_hostname,
			inode->i_sb->s_id, (long long)nfsi->fileid,
			(long long)fattr->fileid);
		goto out_err;
	}
L
Linus Torvalds 已提交
1834 1835 1836 1837

	/*
	 * Make sure the inode's type hasn't changed.
	 */
1838 1839 1840 1841
	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT)) {
		/*
		* Big trouble! The inode has become a different object.
		*/
1842
		printk(KERN_DEBUG "NFS: %s: inode %lu mode changed, %07o to %07o\n",
1843 1844 1845
				__func__, inode->i_ino, inode->i_mode, fattr->mode);
		goto out_err;
	}
L
Linus Torvalds 已提交
1846

1847
	server = NFS_SERVER(inode);
T
Trond Myklebust 已提交
1848
	/* Update the fsid? */
1849
	if (S_ISDIR(inode->i_mode) && (fattr->valid & NFS_ATTR_FATTR_FSID) &&
1850
			!nfs_fsid_equal(&server->fsid, &fattr->fsid) &&
1851
			!IS_AUTOMOUNT(inode))
1852 1853
		server->fsid = fattr->fsid;

1854 1855 1856
	/* Save the delegation state before clearing cache_validity */
	have_delegation = nfs_have_delegated_attributes(inode);

L
Linus Torvalds 已提交
1857 1858 1859
	/*
	 * Update the read time so we don't revalidate too often.
	 */
1860
	nfsi->read_cache_jiffies = fattr->time_start;
1861

1862 1863 1864 1865 1866
	save_cache_validity = nfsi->cache_validity;
	nfsi->cache_validity &= ~(NFS_INO_INVALID_ATTR
			| NFS_INO_INVALID_ATIME
			| NFS_INO_REVAL_FORCED
			| NFS_INO_REVAL_PAGECACHE);
L
Linus Torvalds 已提交
1867

1868
	/* Do atomic weak cache consistency updates */
1869
	nfs_wcc_update_inode(inode, fattr);
1870

1871 1872 1873 1874
	if (pnfs_layoutcommit_outstanding(inode)) {
		nfsi->cache_validity |= save_cache_validity & NFS_INO_INVALID_ATTR;
		cache_revalidated = false;
	}
1875

1876
	/* More cache consistency checks */
1877
	if (fattr->valid & NFS_ATTR_FATTR_CHANGE) {
1878
		if (!inode_eq_iversion_raw(inode, fattr->change_attr)) {
1879
			/* Could it be a race with writeback? */
1880
			if (!(have_writers || have_delegation)) {
1881
				invalid |= NFS_INO_INVALID_DATA
1882 1883
					| NFS_INO_INVALID_ACCESS
					| NFS_INO_INVALID_ACL;
1884
				/* Force revalidate of all attributes */
1885 1886 1887 1888
				save_cache_validity |= NFS_INO_INVALID_CTIME
					| NFS_INO_INVALID_MTIME
					| NFS_INO_INVALID_SIZE
					| NFS_INO_INVALID_OTHER;
1889 1890
				if (S_ISDIR(inode->i_mode))
					nfs_force_lookup_revalidate(inode);
1891 1892 1893
				dprintk("NFS: change_attr change on server for file %s/%ld\n",
						inode->i_sb->s_id,
						inode->i_ino);
1894 1895
			} else if (!have_delegation)
				nfsi->cache_validity |= NFS_INO_DATA_INVAL_DEFER;
J
Jeff Layton 已提交
1896
			inode_set_iversion_raw(inode, fattr->change_attr);
1897
			attr_changed = true;
1898
		}
1899
	} else {
1900
		nfsi->cache_validity |= save_cache_validity &
1901
				(NFS_INO_INVALID_CHANGE
1902 1903
				| NFS_INO_REVAL_PAGECACHE
				| NFS_INO_REVAL_FORCED);
1904 1905
		cache_revalidated = false;
	}
1906 1907

	if (fattr->valid & NFS_ATTR_FATTR_MTIME) {
1908
		inode->i_mtime = fattr->mtime;
1909
	} else if (server->caps & NFS_CAP_MTIME) {
1910
		nfsi->cache_validity |= save_cache_validity &
1911
				(NFS_INO_INVALID_MTIME
1912
				| NFS_INO_REVAL_FORCED);
1913 1914
		cache_revalidated = false;
	}
1915

1916
	if (fattr->valid & NFS_ATTR_FATTR_CTIME) {
1917
		inode->i_ctime = fattr->ctime;
1918
	} else if (server->caps & NFS_CAP_CTIME) {
1919
		nfsi->cache_validity |= save_cache_validity &
1920
				(NFS_INO_INVALID_CTIME
1921
				| NFS_INO_REVAL_FORCED);
1922 1923
		cache_revalidated = false;
	}
1924

1925
	/* Check if our cached file size is stale */
1926 1927 1928
	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
		new_isize = nfs_size_to_loff_t(fattr->size);
		cur_isize = i_size_read(inode);
1929
		if (new_isize != cur_isize && !have_delegation) {
1930 1931
			/* Do we perhaps have any outstanding writes, or has
			 * the file grown beyond our last write? */
1932
			if (!nfs_have_writebacks(inode) || new_isize > cur_isize) {
1933
				i_size_write(inode, new_isize);
1934
				if (!have_writers)
1935
					invalid |= NFS_INO_INVALID_DATA;
1936
				attr_changed = true;
1937
			}
1938 1939 1940 1941 1942 1943
			dprintk("NFS: isize change on server for file %s/%ld "
					"(%Ld to %Ld)\n",
					inode->i_sb->s_id,
					inode->i_ino,
					(long long)cur_isize,
					(long long)new_isize);
L
Linus Torvalds 已提交
1944
		}
1945
	} else {
1946
		nfsi->cache_validity |= save_cache_validity &
1947
				(NFS_INO_INVALID_SIZE
1948 1949
				| NFS_INO_REVAL_PAGECACHE
				| NFS_INO_REVAL_FORCED);
1950 1951
		cache_revalidated = false;
	}
L
Linus Torvalds 已提交
1952 1953


1954
	if (fattr->valid & NFS_ATTR_FATTR_ATIME)
1955
		inode->i_atime = fattr->atime;
1956
	else if (server->caps & NFS_CAP_ATIME) {
1957 1958
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATIME
1959
				| NFS_INO_REVAL_FORCED);
1960 1961
		cache_revalidated = false;
	}
L
Linus Torvalds 已提交
1962

1963 1964
	if (fattr->valid & NFS_ATTR_FATTR_MODE) {
		if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) {
1965 1966 1967
			umode_t newmode = inode->i_mode & S_IFMT;
			newmode |= fattr->mode & S_IALLUGO;
			inode->i_mode = newmode;
1968
			invalid |= NFS_INO_INVALID_ACCESS
1969 1970
				| NFS_INO_INVALID_ACL;
			attr_changed = true;
1971
		}
1972
	} else if (server->caps & NFS_CAP_MODE) {
1973
		nfsi->cache_validity |= save_cache_validity &
1974
				(NFS_INO_INVALID_OTHER
1975
				| NFS_INO_REVAL_FORCED);
1976 1977
		cache_revalidated = false;
	}
1978

1979
	if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
1980
		if (!uid_eq(inode->i_uid, fattr->uid)) {
1981
			invalid |= NFS_INO_INVALID_ACCESS
1982
				| NFS_INO_INVALID_ACL;
1983
			inode->i_uid = fattr->uid;
1984
			attr_changed = true;
1985
		}
1986
	} else if (server->caps & NFS_CAP_OWNER) {
1987
		nfsi->cache_validity |= save_cache_validity &
1988
				(NFS_INO_INVALID_OTHER
1989
				| NFS_INO_REVAL_FORCED);
1990 1991
		cache_revalidated = false;
	}
1992

1993
	if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
1994
		if (!gid_eq(inode->i_gid, fattr->gid)) {
1995
			invalid |= NFS_INO_INVALID_ACCESS
1996
				| NFS_INO_INVALID_ACL;
1997
			inode->i_gid = fattr->gid;
1998
			attr_changed = true;
1999
		}
2000
	} else if (server->caps & NFS_CAP_OWNER_GROUP) {
2001
		nfsi->cache_validity |= save_cache_validity &
2002
				(NFS_INO_INVALID_OTHER
2003
				| NFS_INO_REVAL_FORCED);
2004 2005
		cache_revalidated = false;
	}
2006

2007 2008 2009 2010
	if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
		if (inode->i_nlink != fattr->nlink) {
			if (S_ISDIR(inode->i_mode))
				invalid |= NFS_INO_INVALID_DATA;
M
Miklos Szeredi 已提交
2011
			set_nlink(inode, fattr->nlink);
2012
			attr_changed = true;
2013
		}
2014
	} else if (server->caps & NFS_CAP_NLINK) {
2015
		nfsi->cache_validity |= save_cache_validity &
2016
				(NFS_INO_INVALID_OTHER
2017
				| NFS_INO_REVAL_FORCED);
2018 2019
		cache_revalidated = false;
	}
L
Linus Torvalds 已提交
2020

2021
	if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
L
Linus Torvalds 已提交
2022 2023 2024 2025
		/*
		 * report the blocks in 512byte units
		 */
		inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
2026
	} else if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
2027
		inode->i_blocks = fattr->du.nfs2.blocks;
2028 2029
	else
		cache_revalidated = false;
L
Linus Torvalds 已提交
2030 2031

	/* Update attrtimeo value if we're out of the unstable period */
2032
	if (attr_changed) {
2033
		invalid &= ~NFS_INO_INVALID_ATTR;
C
Chuck Lever 已提交
2034
		nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE);
L
Linus Torvalds 已提交
2035
		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
2036
		nfsi->attrtimeo_timestamp = now;
2037
		/* Set barrier to be more recent than all outstanding updates */
2038
		nfsi->attr_gencount = nfs_inc_attr_generation_counter();
2039
	} else {
2040 2041 2042 2043 2044 2045 2046
		if (cache_revalidated) {
			if (!time_in_range_open(now, nfsi->attrtimeo_timestamp,
				nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
				nfsi->attrtimeo <<= 1;
				if (nfsi->attrtimeo > NFS_MAXATTRTIMEO(inode))
					nfsi->attrtimeo = NFS_MAXATTRTIMEO(inode);
			}
2047 2048
			nfsi->attrtimeo_timestamp = now;
		}
2049 2050 2051
		/* Set the barrier to be more recent than this fattr */
		if ((long)fattr->gencount - (long)nfsi->attr_gencount > 0)
			nfsi->attr_gencount = fattr->gencount;
L
Linus Torvalds 已提交
2052
	}
2053

L
Linus Torvalds 已提交
2054 2055 2056 2057
	/* Don't invalidate the data if we were to blame */
	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)
				|| S_ISLNK(inode->i_mode)))
		invalid &= ~NFS_INO_INVALID_DATA;
2058
	nfs_set_cache_invalid(inode, invalid);
D
David Howells 已提交
2059

L
Linus Torvalds 已提交
2060
	return 0;
2061
 out_err:
L
Linus Torvalds 已提交
2062 2063 2064 2065 2066 2067 2068 2069 2070
	/*
	 * No need to worry about unhashing the dentry, as the
	 * lookup validation will know that the inode is bad.
	 * (But we fall through to invalidate the caches.)
	 */
	nfs_invalidate_inode(inode);
	return -ESTALE;
}

D
David Howells 已提交
2071
struct inode *nfs_alloc_inode(struct super_block *sb)
L
Linus Torvalds 已提交
2072 2073
{
	struct nfs_inode *nfsi;
F
Firo Yang 已提交
2074
	nfsi = kmem_cache_alloc(nfs_inode_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
2075 2076
	if (!nfsi)
		return NULL;
2077 2078
	nfsi->flags = 0UL;
	nfsi->cache_validity = 0UL;
B
Bryan Schumaker 已提交
2079
#if IS_ENABLED(CONFIG_NFS_V4)
2080 2081
	nfsi->nfs4_acl = NULL;
#endif /* CONFIG_NFS_V4 */
L
Linus Torvalds 已提交
2082 2083
	return &nfsi->vfs_inode;
}
B
Bryan Schumaker 已提交
2084
EXPORT_SYMBOL_GPL(nfs_alloc_inode);
L
Linus Torvalds 已提交
2085

A
Al Viro 已提交
2086
void nfs_free_inode(struct inode *inode)
L
Linus Torvalds 已提交
2087 2088 2089
{
	kmem_cache_free(nfs_inode_cachep, NFS_I(inode));
}
A
Al Viro 已提交
2090
EXPORT_SYMBOL_GPL(nfs_free_inode);
N
Nick Piggin 已提交
2091

A
Andrew Morton 已提交
2092 2093
static inline void nfs4_init_once(struct nfs_inode *nfsi)
{
B
Bryan Schumaker 已提交
2094
#if IS_ENABLED(CONFIG_NFS_V4)
A
Andrew Morton 已提交
2095 2096 2097
	INIT_LIST_HEAD(&nfsi->open_states);
	nfsi->delegation = NULL;
	init_rwsem(&nfsi->rwsem);
2098
	nfsi->layout = NULL;
A
Andrew Morton 已提交
2099 2100
#endif
}
D
David Howells 已提交
2101

2102
static void init_once(void *foo)
L
Linus Torvalds 已提交
2103 2104 2105
{
	struct nfs_inode *nfsi = (struct nfs_inode *) foo;

C
Christoph Lameter 已提交
2106 2107 2108 2109
	inode_init_once(&nfsi->vfs_inode);
	INIT_LIST_HEAD(&nfsi->open_files);
	INIT_LIST_HEAD(&nfsi->access_cache_entry_lru);
	INIT_LIST_HEAD(&nfsi->access_cache_inode_lru);
F
Fred Isaman 已提交
2110
	INIT_LIST_HEAD(&nfsi->commit_info.list);
2111
	atomic_long_set(&nfsi->nrequests, 0);
2112
	atomic_long_set(&nfsi->commit_info.ncommit, 0);
2113
	atomic_set(&nfsi->commit_info.rpcs_out, 0);
A
Al Viro 已提交
2114
	init_rwsem(&nfsi->rmdir_sem);
2115
	mutex_init(&nfsi->commit_mutex);
C
Christoph Lameter 已提交
2116
	nfs4_init_once(nfsi);
2117
	nfsi->cache_change_attribute = 0;
L
Linus Torvalds 已提交
2118
}
2119

D
David Howells 已提交
2120
static int __init nfs_init_inodecache(void)
L
Linus Torvalds 已提交
2121 2122 2123
{
	nfs_inode_cachep = kmem_cache_create("nfs_inode_cache",
					     sizeof(struct nfs_inode),
2124
					     0, (SLAB_RECLAIM_ACCOUNT|
2125
						SLAB_MEM_SPREAD|SLAB_ACCOUNT),
2126
					     init_once);
L
Linus Torvalds 已提交
2127 2128 2129 2130 2131 2132
	if (nfs_inode_cachep == NULL)
		return -ENOMEM;

	return 0;
}

2133
static void nfs_destroy_inodecache(void)
L
Linus Torvalds 已提交
2134
{
2135 2136 2137 2138 2139
	/*
	 * Make sure all delayed rcu free inodes are flushed before we
	 * destroy cache.
	 */
	rcu_barrier();
2140
	kmem_cache_destroy(nfs_inode_cachep);
L
Linus Torvalds 已提交
2141 2142
}

T
Trond Myklebust 已提交
2143
struct workqueue_struct *nfsiod_workqueue;
B
Bryan Schumaker 已提交
2144
EXPORT_SYMBOL_GPL(nfsiod_workqueue);
T
Trond Myklebust 已提交
2145 2146 2147 2148 2149 2150 2151 2152

/*
 * start up the nfsiod workqueue
 */
static int nfsiod_start(void)
{
	struct workqueue_struct *wq;
	dprintk("RPC:       creating workqueue nfsiod\n");
2153
	wq = alloc_workqueue("nfsiod", WQ_MEM_RECLAIM, 0);
T
Trond Myklebust 已提交
2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173
	if (wq == NULL)
		return -ENOMEM;
	nfsiod_workqueue = wq;
	return 0;
}

/*
 * Destroy the nfsiod workqueue
 */
static void nfsiod_stop(void)
{
	struct workqueue_struct *wq;

	wq = nfsiod_workqueue;
	if (wq == NULL)
		return;
	nfsiod_workqueue = NULL;
	destroy_workqueue(wq);
}

2174
unsigned int nfs_net_id;
2175
EXPORT_SYMBOL_GPL(nfs_net_id);
2176 2177 2178

static int nfs_net_init(struct net *net)
{
2179
	nfs_clients_init(net);
2180
	return nfs_fs_proc_net_init(net);
2181 2182 2183 2184
}

static void nfs_net_exit(struct net *net)
{
2185
	nfs_fs_proc_net_exit(net);
2186
	nfs_clients_exit(net);
2187 2188 2189 2190 2191 2192 2193 2194 2195
}

static struct pernet_operations nfs_net_ops = {
	.init = nfs_net_init,
	.exit = nfs_net_exit,
	.id   = &nfs_net_id,
	.size = sizeof(struct nfs_net),
};

L
Linus Torvalds 已提交
2196 2197 2198 2199 2200 2201 2202
/*
 * Initialize NFS
 */
static int __init init_nfs_fs(void)
{
	int err;

2203 2204 2205 2206
	err = nfs_sysfs_init();
	if (err < 0)
		goto out10;

2207
	err = register_pernet_subsys(&nfs_net_ops);
2208
	if (err < 0)
B
Bryan Schumaker 已提交
2209
		goto out9;
2210

2211 2212
	err = nfs_fscache_register();
	if (err < 0)
B
Bryan Schumaker 已提交
2213
		goto out8;
2214

T
Trond Myklebust 已提交
2215 2216
	err = nfsiod_start();
	if (err)
B
Bryan Schumaker 已提交
2217
		goto out7;
T
Trond Myklebust 已提交
2218

2219 2220
	err = nfs_fs_proc_init();
	if (err)
B
Bryan Schumaker 已提交
2221
		goto out6;
2222

L
Linus Torvalds 已提交
2223 2224
	err = nfs_init_nfspagecache();
	if (err)
B
Bryan Schumaker 已提交
2225
		goto out5;
L
Linus Torvalds 已提交
2226 2227 2228

	err = nfs_init_inodecache();
	if (err)
B
Bryan Schumaker 已提交
2229
		goto out4;
L
Linus Torvalds 已提交
2230 2231 2232

	err = nfs_init_readpagecache();
	if (err)
B
Bryan Schumaker 已提交
2233
		goto out3;
L
Linus Torvalds 已提交
2234 2235 2236

	err = nfs_init_writepagecache();
	if (err)
B
Bryan Schumaker 已提交
2237
		goto out2;
L
Linus Torvalds 已提交
2238 2239 2240

	err = nfs_init_directcache();
	if (err)
B
Bryan Schumaker 已提交
2241
		goto out1;
L
Linus Torvalds 已提交
2242

2243
	rpc_proc_register(&init_net, &nfs_rpcstat);
2244 2245 2246

	err = register_nfs_fs();
	if (err)
2247 2248
		goto out0;

L
Linus Torvalds 已提交
2249
	return 0;
2250
out0:
2251
	rpc_proc_unregister(&init_net, "nfs");
L
Linus Torvalds 已提交
2252
	nfs_destroy_directcache();
B
Bryan Schumaker 已提交
2253
out1:
2254
	nfs_destroy_writepagecache();
B
Bryan Schumaker 已提交
2255
out2:
2256
	nfs_destroy_readpagecache();
B
Bryan Schumaker 已提交
2257
out3:
2258
	nfs_destroy_inodecache();
B
Bryan Schumaker 已提交
2259
out4:
2260
	nfs_destroy_nfspagecache();
B
Bryan Schumaker 已提交
2261
out5:
2262
	nfs_fs_proc_exit();
B
Bryan Schumaker 已提交
2263
out6:
2264
	nfsiod_stop();
B
Bryan Schumaker 已提交
2265
out7:
2266
	nfs_fscache_unregister();
B
Bryan Schumaker 已提交
2267
out8:
2268
	unregister_pernet_subsys(&nfs_net_ops);
B
Bryan Schumaker 已提交
2269
out9:
2270 2271
	nfs_sysfs_exit();
out10:
L
Linus Torvalds 已提交
2272 2273 2274 2275 2276 2277 2278 2279 2280 2281
	return err;
}

static void __exit exit_nfs_fs(void)
{
	nfs_destroy_directcache();
	nfs_destroy_writepagecache();
	nfs_destroy_readpagecache();
	nfs_destroy_inodecache();
	nfs_destroy_nfspagecache();
2282
	nfs_fscache_unregister();
2283
	unregister_pernet_subsys(&nfs_net_ops);
2284
	rpc_proc_unregister(&init_net, "nfs");
D
David Howells 已提交
2285
	unregister_nfs_fs();
2286
	nfs_fs_proc_exit();
T
Trond Myklebust 已提交
2287
	nfsiod_stop();
2288
	nfs_sysfs_exit();
L
Linus Torvalds 已提交
2289 2290 2291 2292 2293
}

/* Not quite true; I just maintain it */
MODULE_AUTHOR("Olaf Kirch <okir@monad.swb.de>");
MODULE_LICENSE("GPL");
2294
module_param(enable_ino64, bool, 0644);
L
Linus Torvalds 已提交
2295 2296 2297

module_init(init_nfs_fs)
module_exit(exit_nfs_fs)