inode.c 56.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 *  linux/fs/nfs/inode.c
 *
 *  Copyright (C) 1992  Rick Sladkey
 *
 *  nfs inode and superblock handling functions
 *
A
Alan Cox 已提交
8
 *  Modularised by Alan Cox <alan@lxorguk.ukuu.org.uk>, while hacking some
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16 17
 *  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>
A
Alexey Dobriyan 已提交
18
#include <linux/sched.h>
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26 27
#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>
28
#include <linux/sunrpc/metrics.h>
L
Linus Torvalds 已提交
29 30 31 32 33 34 35
#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>
36 37
#include <linux/inet.h>
#include <linux/nfs_xdr.h>
38
#include <linux/slab.h>
39
#include <linux/compat.h>
40
#include <linux/freezer.h>
L
Linus Torvalds 已提交
41 42 43

#include <asm/uaccess.h>

44
#include "nfs4_fs.h"
45
#include "callback.h"
L
Linus Torvalds 已提交
46
#include "delegation.h"
47
#include "iostat.h"
D
David Howells 已提交
48
#include "internal.h"
49
#include "fscache.h"
50
#include "pnfs.h"
51
#include "nfs.h"
52
#include "netns.h"
L
Linus Torvalds 已提交
53

54 55
#include "nfstrace.h"

L
Linus Torvalds 已提交
56 57
#define NFSDBG_FACILITY		NFSDBG_VFS

58 59 60
#define NFS_64_BIT_INODE_NUMBERS_ENABLED	1

/* Default is to see 64-bit inode numbers */
61
static bool enable_ino64 = NFS_64_BIT_INODE_NUMBERS_ENABLED;
62

L
Linus Torvalds 已提交
63
static void nfs_invalidate_inode(struct inode *);
64
static int nfs_update_inode(struct inode *, struct nfs_fattr *);
L
Linus Torvalds 已提交
65

66
static struct kmem_cache * nfs_inode_cachep;
67

L
Linus Torvalds 已提交
68 69 70 71 72 73
static inline unsigned long
nfs_fattr_to_ino_t(struct nfs_fattr *fattr)
{
	return nfs_fileid_to_ino_t(fattr->fileid);
}

74 75 76 77
/**
 * nfs_wait_bit_killable - helper for functions that are sleeping on bit locks
 * @word: long word containing the bit lock
 */
78
int nfs_wait_bit_killable(struct wait_bit_key *key)
79 80 81
{
	if (fatal_signal_pending(current))
		return -ERESTARTSYS;
82
	freezable_schedule_unsafe();
83 84
	return 0;
}
B
Bryan Schumaker 已提交
85
EXPORT_SYMBOL_GPL(nfs_wait_bit_killable);
86

87 88 89 90 91 92 93 94 95
/**
 * 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)
{
96 97 98 99 100
#ifdef CONFIG_COMPAT
	compat_ulong_t ino;
#else	
	unsigned long ino;
#endif
101 102 103 104 105 106 107 108 109

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

110 111 112 113 114 115
int nfs_drop_inode(struct inode *inode)
{
	return NFS_STALE(inode) || generic_drop_inode(inode);
}
EXPORT_SYMBOL_GPL(nfs_drop_inode);

116
void nfs_clear_inode(struct inode *inode)
L
Linus Torvalds 已提交
117
{
T
Trond Myklebust 已提交
118 119 120
	/*
	 * The following should never happen...
	 */
121 122
	WARN_ON_ONCE(nfs_have_writebacks(inode));
	WARN_ON_ONCE(!list_empty(&NFS_I(inode)->open_files));
123
	nfs_zap_acl_cache(inode);
124
	nfs_access_zap_cache(inode);
125
	nfs_fscache_clear_inode(inode);
L
Linus Torvalds 已提交
126
}
B
Bryan Schumaker 已提交
127
EXPORT_SYMBOL_GPL(nfs_clear_inode);
L
Linus Torvalds 已提交
128

129 130
void nfs_evict_inode(struct inode *inode)
{
131
	truncate_inode_pages_final(&inode->i_data);
132
	clear_inode(inode);
133 134 135
	nfs_clear_inode(inode);
}

136
int nfs_sync_inode(struct inode *inode)
137 138 139 140
{
	nfs_inode_dio_wait(inode);
	return nfs_wb_all(inode);
}
141
EXPORT_SYMBOL_GPL(nfs_sync_inode);
142

T
Trond Myklebust 已提交
143 144 145 146 147
/**
 * nfs_sync_mapping - helper to flush all mmapped dirty data to disk
 */
int nfs_sync_mapping(struct address_space *mapping)
{
T
Trond Myklebust 已提交
148
	int ret = 0;
T
Trond Myklebust 已提交
149

T
Trond Myklebust 已提交
150 151 152 153
	if (mapping->nrpages != 0) {
		unmap_mapping_range(mapping, 0, 0, 0);
		ret = nfs_wb_all(mapping->host);
	}
T
Trond Myklebust 已提交
154 155 156
	return ret;
}

157 158 159 160 161 162 163 164 165 166 167
static void nfs_set_cache_invalid(struct inode *inode, unsigned long flags)
{
	struct nfs_inode *nfsi = NFS_I(inode);

	if (inode->i_mapping->nrpages == 0)
		flags &= ~NFS_INO_INVALID_DATA;
	nfsi->cache_validity |= flags;
	if (flags & NFS_INO_INVALID_DATA)
		nfs_fscache_invalidate(inode);
}

L
Linus Torvalds 已提交
168 169 170
/*
 * Invalidate the local caches
 */
171
static void nfs_zap_caches_locked(struct inode *inode)
L
Linus Torvalds 已提交
172 173 174 175
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int mode = inode->i_mode;

C
Chuck Lever 已提交
176 177
	nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE);

178 179
	nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
	nfsi->attrtimeo_timestamp = jiffies;
L
Linus Torvalds 已提交
180

181
	memset(NFS_I(inode)->cookieverf, 0, sizeof(NFS_I(inode)->cookieverf));
D
David Howells 已提交
182
	if (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)) {
183
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
184 185 186
					| NFS_INO_INVALID_DATA
					| NFS_INO_INVALID_ACCESS
					| NFS_INO_INVALID_ACL
187
					| NFS_INO_REVAL_PAGECACHE);
188
	} else
189
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
190 191
					| NFS_INO_INVALID_ACCESS
					| NFS_INO_INVALID_ACL
192
					| NFS_INO_REVAL_PAGECACHE);
193
	nfs_zap_label_cache_locked(nfsi);
194
}
195

196 197 198 199
void nfs_zap_caches(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	nfs_zap_caches_locked(inode);
200
	spin_unlock(&inode->i_lock);
201 202
}

203 204 205 206
void nfs_zap_mapping(struct inode *inode, struct address_space *mapping)
{
	if (mapping->nrpages != 0) {
		spin_lock(&inode->i_lock);
207
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
208 209 210 211
		spin_unlock(&inode->i_lock);
	}
}

212
void nfs_zap_acl_cache(struct inode *inode)
213 214 215 216 217 218
{
	void (*clear_acl_cache)(struct inode *);

	clear_acl_cache = NFS_PROTO(inode)->clear_acl_cache;
	if (clear_acl_cache != NULL)
		clear_acl_cache(inode);
219
	spin_lock(&inode->i_lock);
220
	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_ACL;
221
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
222
}
B
Bryan Schumaker 已提交
223
EXPORT_SYMBOL_GPL(nfs_zap_acl_cache);
L
Linus Torvalds 已提交
224

225 226 227
void nfs_invalidate_atime(struct inode *inode)
{
	spin_lock(&inode->i_lock);
228
	nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME);
229 230
	spin_unlock(&inode->i_lock);
}
B
Bryan Schumaker 已提交
231
EXPORT_SYMBOL_GPL(nfs_invalidate_atime);
232

L
Linus Torvalds 已提交
233
/*
234 235
 * Invalidate, but do not unhash, the inode.
 * NB: must be called with inode->i_lock held!
L
Linus Torvalds 已提交
236
 */
237
static void nfs_invalidate_inode(struct inode *inode)
L
Linus Torvalds 已提交
238
{
B
Benny Halevy 已提交
239
	set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
240
	nfs_zap_caches_locked(inode);
L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
}

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;
263 264
	if ((S_IFMT & inode->i_mode) != (S_IFMT & fattr->mode))
		return 0;
L
Linus Torvalds 已提交
265 266 267 268 269 270 271 272 273 274 275 276 277
	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;

278
	set_nfs_fileid(inode, fattr->fileid);
L
Linus Torvalds 已提交
279 280 281 282
	nfs_copy_fh(NFS_FH(inode), desc->fh);
	return 0;
}

283
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
284 285 286 287 288 289 290
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);
}

291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
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);
308
		nfs_clear_label_invalid(inode);
309 310 311
	}
}

312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
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);
337
#else
338
void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
339 340 341
					struct nfs4_label *label)
{
}
342
#endif
343
EXPORT_SYMBOL_GPL(nfs_setsecurity);
344

L
Linus Torvalds 已提交
345 346 347 348 349
/*
 * This is our front-end to iget that looks up inodes by file handle
 * instead of inode number.
 */
struct inode *
350
nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
351 352 353 354 355
{
	struct nfs_find_desc desc = {
		.fh	= fh,
		.fattr	= fattr
	};
356
	struct inode *inode = ERR_PTR(-ENOENT);
L
Linus Torvalds 已提交
357 358
	unsigned long hash;

359 360
	nfs_attr_check_mountpoint(sb, fattr);

361 362 363
	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
364
		goto out_no_inode;
365
	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) == 0)
L
Linus Torvalds 已提交
366 367 368 369
		goto out_no_inode;

	hash = nfs_fattr_to_ino_t(fattr);

370 371 372
	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
373
		goto out_no_inode;
374
	}
L
Linus Torvalds 已提交
375 376 377

	if (inode->i_state & I_NEW) {
		struct nfs_inode *nfsi = NFS_I(inode);
378
		unsigned long now = jiffies;
L
Linus Torvalds 已提交
379 380 381 382 383 384 385 386

		/* 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;
387 388
		if ((fattr->valid & NFS_ATTR_FATTR_MODE) == 0
				&& nfs_server_capable(inode, NFS_CAP_MODE))
389
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
L
Linus Torvalds 已提交
390 391 392
		/* Why so? Because we want revalidate for devices/FIFOs, and
		 * that's precisely what we have in nfs_file_inode_operations.
		 */
393
		inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->file_inode_ops;
L
Linus Torvalds 已提交
394
		if (S_ISREG(inode->i_mode)) {
395
			inode->i_fop = NFS_SB(sb)->nfs_client->rpc_ops->file_ops;
L
Linus Torvalds 已提交
396 397
			inode->i_data.a_ops = &nfs_file_aops;
		} else if (S_ISDIR(inode->i_mode)) {
398
			inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->dir_inode_ops;
L
Linus Torvalds 已提交
399
			inode->i_fop = &nfs_dir_operations;
400
			inode->i_data.a_ops = &nfs_dir_aops;
401
			/* Deal with crossing mountpoints */
402 403
			if (fattr->valid & NFS_ATTR_FATTR_MOUNTPOINT ||
					fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) {
M
Manoj Naik 已提交
404 405 406 407
				if (fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL)
					inode->i_op = &nfs_referral_inode_operations;
				else
					inode->i_op = &nfs_mountpoint_inode_operations;
408
				inode->i_fop = NULL;
409
				inode->i_flags |= S_AUTOMOUNT;
410
			}
L
Linus Torvalds 已提交
411 412 413 414 415
		} else if (S_ISLNK(inode->i_mode))
			inode->i_op = &nfs_symlink_inode_operations;
		else
			init_special_inode(inode, inode->i_mode, fattr->rdev);

416 417 418
		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));
419
		inode->i_version = 0;
420
		inode->i_size = 0;
421
		clear_nlink(inode);
422 423
		inode->i_uid = make_kuid(&init_user_ns, -2);
		inode->i_gid = make_kgid(&init_user_ns, -2);
424 425
		inode->i_blocks = 0;
		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
426 427
		nfsi->write_io = 0;
		nfsi->read_io = 0;
428

429
		nfsi->read_cache_jiffies = fattr->time_start;
430
		nfsi->attr_gencount = fattr->gencount;
431 432
		if (fattr->valid & NFS_ATTR_FATTR_ATIME)
			inode->i_atime = fattr->atime;
433
		else if (nfs_server_capable(inode, NFS_CAP_ATIME))
434
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
435 436
		if (fattr->valid & NFS_ATTR_FATTR_MTIME)
			inode->i_mtime = fattr->mtime;
437
		else if (nfs_server_capable(inode, NFS_CAP_MTIME))
438
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
439 440
		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
			inode->i_ctime = fattr->ctime;
441
		else if (nfs_server_capable(inode, NFS_CAP_CTIME))
442
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
443
		if (fattr->valid & NFS_ATTR_FATTR_CHANGE)
444
			inode->i_version = fattr->change_attr;
445
		else
446 447
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
				| NFS_INO_REVAL_PAGECACHE);
448 449
		if (fattr->valid & NFS_ATTR_FATTR_SIZE)
			inode->i_size = nfs_size_to_loff_t(fattr->size);
450
		else
451 452
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
				| NFS_INO_REVAL_PAGECACHE);
453
		if (fattr->valid & NFS_ATTR_FATTR_NLINK)
M
Miklos Szeredi 已提交
454
			set_nlink(inode, fattr->nlink);
455
		else if (nfs_server_capable(inode, NFS_CAP_NLINK))
456
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
457 458
		if (fattr->valid & NFS_ATTR_FATTR_OWNER)
			inode->i_uid = fattr->uid;
459
		else if (nfs_server_capable(inode, NFS_CAP_OWNER))
460
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
461 462
		if (fattr->valid & NFS_ATTR_FATTR_GROUP)
			inode->i_gid = fattr->gid;
463
		else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP))
464
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR);
465 466 467
		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
468 469 470 471 472
			/*
			 * report the blocks in 512byte units
			 */
			inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
		}
473 474 475

		nfs_setsecurity(inode, fattr, label);

L
Linus Torvalds 已提交
476
		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
477
		nfsi->attrtimeo_timestamp = now;
478
		nfsi->access_cache = RB_ROOT;
L
Linus Torvalds 已提交
479

480
		nfs_fscache_init_inode(inode);
D
David Howells 已提交
481

L
Linus Torvalds 已提交
482 483 484
		unlock_new_inode(inode);
	} else
		nfs_refresh_inode(inode, fattr);
485
	dprintk("NFS: nfs_fhget(%s/%Lu fh_crc=0x%08x ct=%d)\n",
L
Linus Torvalds 已提交
486
		inode->i_sb->s_id,
487
		(unsigned long long)NFS_FILEID(inode),
488
		nfs_display_fhandle_hash(fh),
L
Linus Torvalds 已提交
489 490 491 492 493 494
		atomic_read(&inode->i_count));

out:
	return inode;

out_no_inode:
495
	dprintk("nfs_fhget: iget failed with error %ld\n", PTR_ERR(inode));
L
Linus Torvalds 已提交
496 497
	goto out;
}
B
Bryan Schumaker 已提交
498
EXPORT_SYMBOL_GPL(nfs_fhget);
L
Linus Torvalds 已提交
499

500
#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
501 502 503 504

int
nfs_setattr(struct dentry *dentry, struct iattr *attr)
{
505
	struct inode *inode = d_inode(dentry);
506 507
	struct nfs_fattr *fattr;
	int error = -ENOMEM;
L
Linus Torvalds 已提交
508

C
Chuck Lever 已提交
509 510
	nfs_inc_stats(inode, NFSIOS_VFSSETATTR);

511 512 513 514
	/* 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
515
	if (attr->ia_valid & ATTR_SIZE) {
516 517
		loff_t i_size;

518 519
		BUG_ON(!S_ISREG(inode->i_mode));

520 521
		i_size = i_size_read(inode);
		if (attr->ia_size == i_size)
L
Linus Torvalds 已提交
522
			attr->ia_valid &= ~ATTR_SIZE;
523 524
		else if (attr->ia_size < i_size && IS_SWAPFILE(inode))
			return -ETXTBSY;
L
Linus Torvalds 已提交
525 526 527 528
	}

	/* Optimization: if the end result is no change, don't RPC */
	attr->ia_valid &= NFS_VALID_ATTRS;
529
	if ((attr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
L
Linus Torvalds 已提交
530 531
		return 0;

532 533
	trace_nfs_setattr_enter(inode);

534
	/* Write all dirty data */
535 536
	if (S_ISREG(inode->i_mode))
		nfs_sync_inode(inode);
537 538 539 540

	fattr = nfs_alloc_fattr();
	if (fattr == NULL)
		goto out;
541 542 543 544
	/*
	 * Return any delegations if we're going to change ACLs
	 */
	if ((attr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID)) != 0)
545
		NFS_PROTO(inode)->return_delegation(inode);
546
	error = NFS_PROTO(inode)->setattr(dentry, fattr, attr);
547
	if (error == 0)
548
		error = nfs_refresh_inode(inode, fattr);
549 550
	nfs_free_fattr(fattr);
out:
551
	trace_nfs_setattr_exit(inode, error);
552 553
	return error;
}
B
Bryan Schumaker 已提交
554
EXPORT_SYMBOL_GPL(nfs_setattr);
555

556 557 558 559 560 561 562 563
/**
 * 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.
564
 * Note: must be called with inode->i_lock held!
565 566 567
 */
static int nfs_vmtruncate(struct inode * inode, loff_t offset)
{
N
npiggin@suse.de 已提交
568
	int err;
569

N
npiggin@suse.de 已提交
570 571 572
	err = inode_newsize_ok(inode, offset);
	if (err)
		goto out;
573

N
npiggin@suse.de 已提交
574
	i_size_write(inode, offset);
575 576 577
	/* Optimisation */
	if (offset == 0)
		NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_DATA;
N
npiggin@suse.de 已提交
578

579
	spin_unlock(&inode->i_lock);
580
	truncate_pagecache(inode, offset);
581
	spin_lock(&inode->i_lock);
N
npiggin@suse.de 已提交
582 583
out:
	return err;
584 585
}

586 587 588 589 590 591 592 593
/**
 * nfs_setattr_update_inode - Update inode metadata after a setattr call.
 * @inode: pointer to struct inode
 * @attr: pointer to struct iattr
 *
 * Note: we do this in the *proc.c in order to ensure that
 *       it works for things like exclusive creates too.
 */
594 595
void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr,
		struct nfs_fattr *fattr)
596
{
597 598 599 600 601
	/* Barrier: bump the attribute generation count. */
	nfs_fattr_set_barrier(fattr);

	spin_lock(&inode->i_lock);
	NFS_I(inode)->attr_gencount = fattr->gencount;
602
	if ((attr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID)) != 0) {
L
Linus Torvalds 已提交
603
		if ((attr->ia_valid & ATTR_MODE) != 0) {
604 605
			int mode = attr->ia_mode & S_IALLUGO;
			mode |= inode->i_mode & ~S_IALLUGO;
L
Linus Torvalds 已提交
606 607 608 609 610 611
			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;
612 613
		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ACCESS
				| NFS_INO_INVALID_ACL);
614 615
	}
	if ((attr->ia_valid & ATTR_SIZE) != 0) {
C
Chuck Lever 已提交
616
		nfs_inc_stats(inode, NFSIOS_SETATTRTRUNC);
617
		nfs_vmtruncate(inode, attr->ia_size);
618
	}
619 620
	nfs_update_inode(inode, fattr);
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
621
}
B
Bryan Schumaker 已提交
622
EXPORT_SYMBOL_GPL(nfs_setattr_update_inode);
L
Linus Torvalds 已提交
623

624 625 626 627 628
static void nfs_request_parent_use_readdirplus(struct dentry *dentry)
{
	struct dentry *parent;

	parent = dget_parent(dentry);
629
	nfs_force_use_readdirplus(d_inode(parent));
630 631 632 633 634 635 636 637 638 639 640 641 642
	dput(parent);
}

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;
}

L
Linus Torvalds 已提交
643 644
int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
{
645
	struct inode *inode = d_inode(dentry);
646
	int need_atime = NFS_I(inode)->cache_validity & NFS_INO_INVALID_ATIME;
647
	int err = 0;
L
Linus Torvalds 已提交
648

649
	trace_nfs_getattr_enter(inode);
650
	/* Flush out writes to the server in order to update c/mtime.  */
651
	if (S_ISREG(inode->i_mode)) {
652
		mutex_lock(&inode->i_mutex);
653
		err = nfs_sync_inode(inode);
654
		mutex_unlock(&inode->i_mutex);
655 656
		if (err)
			goto out;
657
	}
658 659 660 661 662 663 664 665 666 667 668 669

	/*
	 * 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
	 *  - NFS never sets MS_NOATIME or MS_NODIRATIME so there is
	 *    no point in checking those.
	 */
 	if ((mnt->mnt_flags & MNT_NOATIME) ||
 	    ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
L
Linus Torvalds 已提交
670
		need_atime = 0;
671

672 673 674 675 676 677 678
	if (need_atime || nfs_need_revalidate_inode(inode)) {
		struct nfs_server *server = NFS_SERVER(inode);

		if (server->caps & NFS_CAP_READDIRPLUS)
			nfs_request_parent_use_readdirplus(dentry);
		err = __nfs_revalidate_inode(server, inode);
	}
P
Peter Staubach 已提交
679
	if (!err) {
L
Linus Torvalds 已提交
680
		generic_fillattr(inode, stat);
681
		stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
682 683
		if (S_ISDIR(inode->i_mode))
			stat->blksize = NFS_SERVER(inode)->dtsize;
P
Peter Staubach 已提交
684
	}
685
out:
686
	trace_nfs_getattr_exit(inode, err);
L
Linus Torvalds 已提交
687 688
	return err;
}
B
Bryan Schumaker 已提交
689
EXPORT_SYMBOL_GPL(nfs_getattr);
L
Linus Torvalds 已提交
690

691 692 693
static void nfs_init_lock_context(struct nfs_lock_context *l_ctx)
{
	atomic_set(&l_ctx->count, 1);
694 695
	l_ctx->lockowner.l_owner = current->files;
	l_ctx->lockowner.l_pid = current->tgid;
696
	INIT_LIST_HEAD(&l_ctx->list);
697
	nfs_iocounter_init(&l_ctx->io_count);
698 699 700 701
}

static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx)
{
702 703
	struct nfs_lock_context *head = &ctx->lock_context;
	struct nfs_lock_context *pos = head;
704

705
	do {
706
		if (pos->lockowner.l_owner != current->files)
707
			continue;
708
		if (pos->lockowner.l_pid != current->tgid)
709 710 711
			continue;
		atomic_inc(&pos->count);
		return pos;
712
	} while ((pos = list_entry(pos->list.next, typeof(*pos), list)) != head);
713 714 715 716 717 718
	return NULL;
}

struct nfs_lock_context *nfs_get_lock_context(struct nfs_open_context *ctx)
{
	struct nfs_lock_context *res, *new = NULL;
719
	struct inode *inode = d_inode(ctx->dentry);
720 721 722 723 724 725 726

	spin_lock(&inode->i_lock);
	res = __nfs_find_lock_context(ctx);
	if (res == NULL) {
		spin_unlock(&inode->i_lock);
		new = kmalloc(sizeof(*new), GFP_KERNEL);
		if (new == NULL)
727
			return ERR_PTR(-ENOMEM);
728 729 730 731 732 733 734 735 736 737 738 739 740 741
		nfs_init_lock_context(new);
		spin_lock(&inode->i_lock);
		res = __nfs_find_lock_context(ctx);
		if (res == NULL) {
			list_add_tail(&new->list, &ctx->lock_context.list);
			new->open_context = ctx;
			res = new;
			new = NULL;
		}
	}
	spin_unlock(&inode->i_lock);
	kfree(new);
	return res;
}
A
Anna Schumaker 已提交
742
EXPORT_SYMBOL_GPL(nfs_get_lock_context);
743 744 745 746

void nfs_put_lock_context(struct nfs_lock_context *l_ctx)
{
	struct nfs_open_context *ctx = l_ctx->open_context;
747
	struct inode *inode = d_inode(ctx->dentry);
748 749 750 751 752 753 754

	if (!atomic_dec_and_lock(&l_ctx->count, &inode->i_lock))
		return;
	list_del(&l_ctx->list);
	spin_unlock(&inode->i_lock);
	kfree(l_ctx);
}
A
Anna Schumaker 已提交
755
EXPORT_SYMBOL_GPL(nfs_put_lock_context);
756

T
Trond Myklebust 已提交
757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773
/**
 * nfs_close_context - Common close_context() routine NFSv2/v3
 * @ctx: pointer to context
 * @is_sync: is this a synchronous close
 *
 * always ensure that the attributes are up to date if we're mounted
 * with close-to-open semantics
 */
void nfs_close_context(struct nfs_open_context *ctx, int is_sync)
{
	struct inode *inode;
	struct nfs_server *server;

	if (!(ctx->mode & FMODE_WRITE))
		return;
	if (!is_sync)
		return;
774
	inode = d_inode(ctx->dentry);
T
Trond Myklebust 已提交
775 776 777 778 779 780 781
	if (!list_empty(&NFS_I(inode)->open_files))
		return;
	server = NFS_SERVER(inode);
	if (server->flags & NFS_MOUNT_NOCTO)
		return;
	nfs_revalidate_inode(server, inode);
}
B
Bryan Schumaker 已提交
782
EXPORT_SYMBOL_GPL(nfs_close_context);
T
Trond Myklebust 已提交
783

784
struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, fmode_t f_mode)
L
Linus Torvalds 已提交
785 786
{
	struct nfs_open_context *ctx;
787 788 789
	struct rpc_cred *cred = rpc_lookup_cred();
	if (IS_ERR(cred))
		return ERR_CAST(cred);
L
Linus Torvalds 已提交
790

791
	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
792 793 794
	if (!ctx) {
		put_rpccred(cred);
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
795
	}
796 797 798 799 800 801 802 803 804 805
	nfs_sb_active(dentry->d_sb);
	ctx->dentry = dget(dentry);
	ctx->cred = cred;
	ctx->state = NULL;
	ctx->mode = f_mode;
	ctx->flags = 0;
	ctx->error = 0;
	nfs_init_lock_context(&ctx->lock_context);
	ctx->lock_context.open_context = ctx;
	INIT_LIST_HEAD(&ctx->list);
806
	ctx->mdsthreshold = NULL;
L
Linus Torvalds 已提交
807 808
	return ctx;
}
B
Bryan Schumaker 已提交
809
EXPORT_SYMBOL_GPL(alloc_nfs_open_context);
L
Linus Torvalds 已提交
810 811 812 813

struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx)
{
	if (ctx != NULL)
814
		atomic_inc(&ctx->lock_context.count);
L
Linus Torvalds 已提交
815 816
	return ctx;
}
B
Bryan Schumaker 已提交
817
EXPORT_SYMBOL_GPL(get_nfs_open_context);
L
Linus Torvalds 已提交
818

T
Trond Myklebust 已提交
819
static void __put_nfs_open_context(struct nfs_open_context *ctx, int is_sync)
L
Linus Torvalds 已提交
820
{
821
	struct inode *inode = d_inode(ctx->dentry);
822
	struct super_block *sb = ctx->dentry->d_sb;
823

824
	if (!list_empty(&ctx->list)) {
825 826 827 828 829
		if (!atomic_dec_and_lock(&ctx->lock_context.count, &inode->i_lock))
			return;
		list_del(&ctx->list);
		spin_unlock(&inode->i_lock);
	} else if (!atomic_dec_and_test(&ctx->lock_context.count))
T
Trond Myklebust 已提交
830
		return;
831 832
	if (inode != NULL)
		NFS_PROTO(inode)->close_context(ctx, is_sync);
833 834
	if (ctx->cred != NULL)
		put_rpccred(ctx->cred);
835
	dput(ctx->dentry);
836
	nfs_sb_deactive(sb);
837
	kfree(ctx->mdsthreshold);
838 839 840
	kfree(ctx);
}

841 842 843 844
void put_nfs_open_context(struct nfs_open_context *ctx)
{
	__put_nfs_open_context(ctx, 0);
}
B
Bryan Schumaker 已提交
845
EXPORT_SYMBOL_GPL(put_nfs_open_context);
846

L
Linus Torvalds 已提交
847 848 849 850
/*
 * Ensure that mmap has a recent RPC credential for use when writing out
 * shared pages
 */
851
void nfs_inode_attach_open_context(struct nfs_open_context *ctx)
L
Linus Torvalds 已提交
852
{
853
	struct inode *inode = d_inode(ctx->dentry);
L
Linus Torvalds 已提交
854 855 856 857 858 859
	struct nfs_inode *nfsi = NFS_I(inode);

	spin_lock(&inode->i_lock);
	list_add(&ctx->list, &nfsi->open_files);
	spin_unlock(&inode->i_lock);
}
860 861 862 863 864 865 866 867
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
868
EXPORT_SYMBOL_GPL(nfs_file_set_open_context);
L
Linus Torvalds 已提交
869

870 871 872
/*
 * Given an inode, search for an open context with the desired characteristics
 */
873
struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, fmode_t mode)
L
Linus Torvalds 已提交
874 875 876 877 878 879
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_open_context *pos, *ctx = NULL;

	spin_lock(&inode->i_lock);
	list_for_each_entry(pos, &nfsi->open_files, list) {
880 881
		if (cred != NULL && pos->cred != cred)
			continue;
882 883 884 885
		if ((pos->mode & (FMODE_READ|FMODE_WRITE)) != mode)
			continue;
		ctx = get_nfs_open_context(pos);
		break;
L
Linus Torvalds 已提交
886 887 888 889 890
	}
	spin_unlock(&inode->i_lock);
	return ctx;
}

A
Anna Schumaker 已提交
891
void nfs_file_clear_open_context(struct file *filp)
L
Linus Torvalds 已提交
892
{
893
	struct nfs_open_context *ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
894 895

	if (ctx) {
896
		struct inode *inode = d_inode(ctx->dentry);
897

L
Linus Torvalds 已提交
898 899 900 901
		filp->private_data = NULL;
		spin_lock(&inode->i_lock);
		list_move_tail(&ctx->list, &NFS_I(inode)->open_files);
		spin_unlock(&inode->i_lock);
902
		__put_nfs_open_context(ctx, filp->f_flags & O_DIRECT ? 0 : 1);
L
Linus Torvalds 已提交
903 904 905 906 907 908 909 910 911 912
	}
}

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

913 914 915
	ctx = alloc_nfs_open_context(filp->f_path.dentry, filp->f_mode);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
L
Linus Torvalds 已提交
916 917
	nfs_file_set_open_context(filp, ctx);
	put_nfs_open_context(ctx);
918
	nfs_fscache_open_file(inode, filp);
L
Linus Torvalds 已提交
919 920 921 922 923 924 925 926 927 928 929
	return 0;
}

/*
 * 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;
930
	struct nfs4_label *label = NULL;
931
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
932 933
	struct nfs_inode *nfsi = NFS_I(inode);

934 935
	dfprintk(PAGECACHE, "NFS: revalidating (%s/%Lu)\n",
		inode->i_sb->s_id, (unsigned long long)NFS_FILEID(inode));
L
Linus Torvalds 已提交
936

937 938
	trace_nfs_revalidate_inode_enter(inode);

939
	if (is_bad_inode(inode))
940
		goto out;
L
Linus Torvalds 已提交
941
	if (NFS_STALE(inode))
942
		goto out;
943

944 945 946 947 948
	status = -ENOMEM;
	fattr = nfs_alloc_fattr();
	if (fattr == NULL)
		goto out;

949
	nfs_inc_stats(inode, NFSIOS_INODEREVALIDATE);
950 951 952 953 954 955 956

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

957
	status = NFS_PROTO(inode)->getattr(server, NFS_FH(inode), fattr, label);
L
Linus Torvalds 已提交
958
	if (status != 0) {
959
		dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) getattr failed, error=%d\n",
L
Linus Torvalds 已提交
960
			 inode->i_sb->s_id,
961
			 (unsigned long long)NFS_FILEID(inode), status);
L
Linus Torvalds 已提交
962 963 964
		if (status == -ESTALE) {
			nfs_zap_caches(inode);
			if (!S_ISDIR(inode->i_mode))
B
Benny Halevy 已提交
965
				set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
966
		}
967
		goto err_out;
L
Linus Torvalds 已提交
968 969
	}

970
	status = nfs_refresh_inode(inode, fattr);
L
Linus Torvalds 已提交
971
	if (status) {
972
		dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) refresh failed, error=%d\n",
L
Linus Torvalds 已提交
973
			 inode->i_sb->s_id,
974
			 (unsigned long long)NFS_FILEID(inode), status);
975
		goto err_out;
L
Linus Torvalds 已提交
976
	}
977

978
	if (nfsi->cache_validity & NFS_INO_INVALID_ACL)
979
		nfs_zap_acl_cache(inode);
980

981 982
	nfs_setsecurity(inode, fattr, label);

983
	dfprintk(PAGECACHE, "NFS: (%s/%Lu) revalidation complete\n",
L
Linus Torvalds 已提交
984
		inode->i_sb->s_id,
985
		(unsigned long long)NFS_FILEID(inode));
L
Linus Torvalds 已提交
986

987 988 989
err_out:
	nfs4_label_free(label);
out:
990
	nfs_free_fattr(fattr);
991
	trace_nfs_revalidate_inode_exit(inode, status);
L
Linus Torvalds 已提交
992 993 994 995 996 997 998
	return status;
}

int nfs_attribute_timeout(struct inode *inode)
{
	struct nfs_inode *nfsi = NFS_I(inode);

999 1000 1001
	return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
}

1002
int nfs_attribute_cache_expired(struct inode *inode)
1003
{
1004
	if (nfs_have_delegated_attributes(inode))
L
Linus Torvalds 已提交
1005
		return 0;
1006
	return nfs_attribute_timeout(inode);
L
Linus Torvalds 已提交
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
}

/**
 * nfs_revalidate_inode - Revalidate the inode attributes
 * @server - pointer to nfs_server struct
 * @inode - pointer to inode struct
 *
 * Updates inode attribute information by retrieving the data from the server.
 */
int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
{
1018
	if (!nfs_need_revalidate_inode(inode))
L
Linus Torvalds 已提交
1019 1020 1021
		return NFS_STALE(inode) ? -ESTALE : 0;
	return __nfs_revalidate_inode(server, inode);
}
B
Bryan Schumaker 已提交
1022
EXPORT_SYMBOL_GPL(nfs_revalidate_inode);
L
Linus Torvalds 已提交
1023

1024 1025 1026 1027 1028 1029 1030 1031 1032
int nfs_revalidate_inode_rcu(struct nfs_server *server, struct inode *inode)
{
	if (!(NFS_I(inode)->cache_validity &
			(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL))
			&& !nfs_attribute_cache_expired(inode))
		return NFS_STALE(inode) ? -ESTALE : 0;
	return -ECHILD;
}

1033
static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping)
1034 1035
{
	struct nfs_inode *nfsi = NFS_I(inode);
1036 1037
	int ret;

1038
	if (mapping->nrpages != 0) {
1039
		if (S_ISREG(inode->i_mode)) {
1040
			unmap_mapping_range(mapping, 0, 0, 0);
1041 1042 1043 1044 1045
			ret = nfs_sync_mapping(mapping);
			if (ret < 0)
				return ret;
		}
		ret = invalidate_inode_pages2(mapping);
1046 1047 1048
		if (ret < 0)
			return ret;
	}
1049 1050
	if (S_ISDIR(inode->i_mode)) {
		spin_lock(&inode->i_lock);
1051
		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
1052 1053
		spin_unlock(&inode->i_lock);
	}
1054
	nfs_inc_stats(inode, NFSIOS_DATAINVALIDATE);
D
David Howells 已提交
1055
	nfs_fscache_wait_on_invalidate(inode);
1056

1057 1058 1059
	dfprintk(PAGECACHE, "NFS: (%s/%Lu) data cache invalidated\n",
			inode->i_sb->s_id,
			(unsigned long long)NFS_FILEID(inode));
1060 1061 1062
	return 0;
}

1063 1064 1065 1066 1067 1068 1069 1070 1071
static bool nfs_mapping_need_revalidate_inode(struct inode *inode)
{
	if (nfs_have_delegated_attributes(inode))
		return false;
	return (NFS_I(inode)->cache_validity & NFS_INO_REVAL_PAGECACHE)
		|| nfs_attribute_timeout(inode)
		|| NFS_STALE(inode);
}

1072
/**
1073
 * __nfs_revalidate_mapping - Revalidate the pagecache
1074 1075
 * @inode - pointer to host inode
 * @mapping - pointer to mapping
1076
 * @may_lock - take inode->i_mutex?
1077
 */
1078 1079 1080
static int __nfs_revalidate_mapping(struct inode *inode,
		struct address_space *mapping,
		bool may_lock)
1081 1082
{
	struct nfs_inode *nfsi = NFS_I(inode);
1083
	unsigned long *bitlock = &nfsi->flags;
1084
	int ret = 0;
1085

1086 1087 1088 1089
	/* swapfiles are not supposed to be shared. */
	if (IS_SWAPFILE(inode))
		goto out;

1090
	if (nfs_mapping_need_revalidate_inode(inode)) {
1091 1092 1093
		ret = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
		if (ret < 0)
			goto out;
1094
	}
1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108

	/*
	 * 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 (;;) {
1109 1110
		ret = wait_on_bit_action(bitlock, NFS_INO_INVALIDATING,
					 nfs_wait_bit_killable, TASK_KILLABLE);
1111 1112
		if (ret)
			goto out;
1113 1114 1115 1116 1117 1118
		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)
1119 1120
			break;
		spin_unlock(&inode->i_lock);
1121
		goto out;
1122 1123
	}

1124
	set_bit(NFS_INO_INVALIDATING, bitlock);
1125
	smp_wmb();
1126 1127 1128
	nfsi->cache_validity &= ~NFS_INO_INVALID_DATA;
	spin_unlock(&inode->i_lock);
	trace_nfs_invalidate_mapping_enter(inode);
1129 1130 1131 1132 1133 1134
	if (may_lock) {
		mutex_lock(&inode->i_mutex);
		ret = nfs_invalidate_mapping(inode, mapping);
		mutex_unlock(&inode->i_mutex);
	} else
		ret = nfs_invalidate_mapping(inode, mapping);
1135 1136
	trace_nfs_invalidate_mapping_exit(inode, ret);

1137
	clear_bit_unlock(NFS_INO_INVALIDATING, bitlock);
1138
	smp_mb__after_atomic();
1139
	wake_up_bit(bitlock, NFS_INO_INVALIDATING);
1140
out:
1141
	return ret;
1142 1143
}

1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
/**
 * nfs_revalidate_mapping - Revalidate the pagecache
 * @inode - pointer to host inode
 * @mapping - pointer to mapping
 */
int nfs_revalidate_mapping(struct inode *inode, struct address_space *mapping)
{
	return __nfs_revalidate_mapping(inode, mapping, false);
}

/**
 * nfs_revalidate_mapping_protected - Revalidate the pagecache
 * @inode - pointer to host inode
 * @mapping - pointer to mapping
 *
 * Differs from nfs_revalidate_mapping() in that it grabs the inode->i_mutex
 * while invalidating the mapping.
 */
int nfs_revalidate_mapping_protected(struct inode *inode, struct address_space *mapping)
{
	return __nfs_revalidate_mapping(inode, mapping, true);
}

1167
static unsigned long nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1168 1169
{
	struct nfs_inode *nfsi = NFS_I(inode);
1170
	unsigned long ret = 0;
1171

1172 1173
	if ((fattr->valid & NFS_ATTR_FATTR_PRECHANGE)
			&& (fattr->valid & NFS_ATTR_FATTR_CHANGE)
1174 1175
			&& inode->i_version == fattr->pre_change_attr) {
		inode->i_version = fattr->change_attr;
1176
		if (S_ISDIR(inode->i_mode))
1177
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
1178
		ret |= NFS_INO_INVALID_ATTR;
1179
	}
1180
	/* If we have atomic WCC data, we may update some attributes */
1181 1182
	if ((fattr->valid & NFS_ATTR_FATTR_PRECTIME)
			&& (fattr->valid & NFS_ATTR_FATTR_CTIME)
1183 1184 1185 1186
			&& timespec_equal(&inode->i_ctime, &fattr->pre_ctime)) {
		memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime));
		ret |= NFS_INO_INVALID_ATTR;
	}
1187 1188 1189 1190

	if ((fattr->valid & NFS_ATTR_FATTR_PREMTIME)
			&& (fattr->valid & NFS_ATTR_FATTR_MTIME)
			&& timespec_equal(&inode->i_mtime, &fattr->pre_mtime)) {
1191 1192
		memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime));
		if (S_ISDIR(inode->i_mode))
1193
			nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
1194
		ret |= NFS_INO_INVALID_ATTR;
1195
	}
1196 1197 1198
	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)
1199
			&& nfsi->nrequests == 0) {
1200 1201 1202
		i_size_write(inode, nfs_size_to_loff_t(fattr->size));
		ret |= NFS_INO_INVALID_ATTR;
	}
D
David Howells 已提交
1203

1204
	return ret;
1205 1206
}

L
Linus Torvalds 已提交
1207
/**
1208
 * nfs_check_inode_attributes - verify consistency of the inode attribute cache
L
Linus Torvalds 已提交
1209 1210 1211 1212 1213 1214 1215
 * @inode - pointer to inode
 * @fattr - updated attributes
 *
 * 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.
 */
1216
static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fattr)
L
Linus Torvalds 已提交
1217 1218 1219
{
	struct nfs_inode *nfsi = NFS_I(inode);
	loff_t cur_size, new_isize;
1220
	unsigned long invalid = 0;
L
Linus Torvalds 已提交
1221

1222

1223 1224
	if (nfs_have_delegated_attributes(inode))
		return 0;
1225
	/* Has the inode gone and changed behind our back? */
1226 1227 1228
	if ((fattr->valid & NFS_ATTR_FATTR_FILEID) && nfsi->fileid != fattr->fileid)
		return -EIO;
	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
1229 1230
		return -EIO;

1231
	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
1232
			inode->i_version != fattr->change_attr)
1233
		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
L
Linus Torvalds 已提交
1234 1235

	/* Verify a few of the more important attributes */
1236
	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec_equal(&inode->i_mtime, &fattr->mtime))
1237
		invalid |= NFS_INO_INVALID_ATTR;
1238

1239 1240 1241
	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
		cur_size = i_size_read(inode);
		new_isize = nfs_size_to_loff_t(fattr->size);
1242
		if (cur_size != new_isize)
1243 1244
			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
	}
1245 1246
	if (nfsi->nrequests != 0)
		invalid &= ~NFS_INO_REVAL_PAGECACHE;
L
Linus Torvalds 已提交
1247 1248

	/* Have any file permissions changed? */
1249 1250
	if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
1251
	if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && !uid_eq(inode->i_uid, fattr->uid))
1252
		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
1253
	if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && !gid_eq(inode->i_gid, fattr->gid))
1254
		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
L
Linus Torvalds 已提交
1255 1256

	/* Has the link count changed? */
1257
	if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink)
1258
		invalid |= NFS_INO_INVALID_ATTR;
L
Linus Torvalds 已提交
1259

1260
	if ((fattr->valid & NFS_ATTR_FATTR_ATIME) && !timespec_equal(&inode->i_atime, &fattr->atime))
1261 1262 1263
		invalid |= NFS_INO_INVALID_ATIME;

	if (invalid != 0)
1264
		nfs_set_cache_invalid(inode, invalid);
L
Linus Torvalds 已提交
1265

1266
	nfsi->read_cache_jiffies = fattr->time_start;
L
Linus Torvalds 已提交
1267 1268 1269
	return 0;
}

1270
static atomic_long_t nfs_attr_generation_counter;
1271 1272 1273

static unsigned long nfs_read_attr_generation_counter(void)
{
1274
	return atomic_long_read(&nfs_attr_generation_counter);
1275 1276 1277 1278
}

unsigned long nfs_inc_attr_generation_counter(void)
{
1279
	return atomic_long_inc_return(&nfs_attr_generation_counter);
1280
}
1281
EXPORT_SYMBOL_GPL(nfs_inc_attr_generation_counter);
1282 1283 1284 1285 1286 1287

void nfs_fattr_init(struct nfs_fattr *fattr)
{
	fattr->valid = 0;
	fattr->time_start = jiffies;
	fattr->gencount = nfs_inc_attr_generation_counter();
1288 1289
	fattr->owner_name = NULL;
	fattr->group_name = NULL;
1290
}
B
Bryan Schumaker 已提交
1291
EXPORT_SYMBOL_GPL(nfs_fattr_init);
1292

1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308
/**
 * 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();
}

1309 1310 1311 1312 1313 1314 1315 1316 1317
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
1318
EXPORT_SYMBOL_GPL(nfs_alloc_fattr);
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328

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
1329
EXPORT_SYMBOL_GPL(nfs_alloc_fhandle);
1330

1331
#ifdef NFS_DEBUG
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
/*
 * _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 */
1344
	return nfs_fhandle_hash(fh);
1345
}
C
Chuck Lever 已提交
1346
EXPORT_SYMBOL_GPL(_nfs_display_fhandle_hash);
1347 1348

/*
1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
 * _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;

1360
	if (fh == NULL || fh->size == 0) {
1361 1362 1363 1364
		printk(KERN_DEFAULT "%s at %p is empty\n", caption, fh);
		return;
	}

1365 1366
	printk(KERN_DEFAULT "%s at %p is %u bytes, crc: 0x%08x:\n",
	       caption, fh, fh->size, _nfs_display_fhandle_hash(fh));
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
	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
1391
EXPORT_SYMBOL_GPL(_nfs_display_fhandle);
1392 1393
#endif

1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
/**
 * nfs_inode_attrs_need_update - check if the inode attributes need updating
 * @inode - pointer to inode
 * @fattr - attributes
 *
 * 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.
1407
 * Note also the check for wraparound of 'attr_gencount'
1408 1409 1410 1411 1412 1413 1414 1415 1416
 *
 * 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);

1417 1418
	return ((long)fattr->gencount - (long)nfsi->attr_gencount) > 0 ||
		((long)nfsi->attr_gencount - (long)nfs_read_attr_generation_counter() > 0);
1419 1420
}

1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
/*
 * Don't trust the change_attribute, mtime, ctime or size if
 * a pnfs LAYOUTCOMMIT is outstanding
 */
static void nfs_inode_attrs_handle_layoutcommit(struct inode *inode,
		struct nfs_fattr *fattr)
{
	if (pnfs_layoutcommit_outstanding(inode))
		fattr->valid &= ~(NFS_ATTR_FATTR_CHANGE |
				NFS_ATTR_FATTR_MTIME |
				NFS_ATTR_FATTR_CTIME |
				NFS_ATTR_FATTR_SIZE);
}

1435 1436
static int nfs_refresh_inode_locked(struct inode *inode, struct nfs_fattr *fattr)
{
1437 1438 1439 1440
	int ret;

	trace_nfs_refresh_inode_enter(inode);

1441 1442
	nfs_inode_attrs_handle_layoutcommit(inode, fattr);

1443
	if (nfs_inode_attrs_need_update(inode, fattr))
1444 1445 1446 1447 1448 1449
		ret = nfs_update_inode(inode, fattr);
	else
		ret = nfs_check_inode_attributes(inode, fattr);

	trace_nfs_refresh_inode_exit(inode, ret);
	return ret;
1450 1451
}

1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
/**
 * nfs_refresh_inode - try to update the inode attribute cache
 * @inode - pointer to inode
 * @fattr - updated attributes
 *
 * 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);
1469
	status = nfs_refresh_inode_locked(inode, fattr);
1470
	spin_unlock(&inode->i_lock);
D
David Howells 已提交
1471

1472 1473
	return status;
}
B
Bryan Schumaker 已提交
1474
EXPORT_SYMBOL_GPL(nfs_refresh_inode);
1475

1476 1477
static int nfs_post_op_update_inode_locked(struct inode *inode, struct nfs_fattr *fattr)
{
1478
	unsigned long invalid = NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
1479

1480 1481 1482 1483 1484 1485 1486
	/*
	 * Don't revalidate the pagecache if we hold a delegation, but do
	 * force an attribute update
	 */
	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
		invalid = NFS_INO_INVALID_ATTR|NFS_INO_REVAL_FORCED;

1487 1488 1489
	if (S_ISDIR(inode->i_mode))
		invalid |= NFS_INO_INVALID_DATA;
	nfs_set_cache_invalid(inode, invalid);
1490 1491 1492 1493 1494
	if ((fattr->valid & NFS_ATTR_FATTR) == 0)
		return 0;
	return nfs_refresh_inode_locked(inode, fattr);
}

1495 1496 1497 1498 1499 1500 1501
/**
 * nfs_post_op_update_inode - try to update the inode attribute cache
 * @inode - pointer to inode
 * @fattr - updated attributes
 *
 * After an operation that has changed the inode metadata, mark the
 * attribute cache as being invalid, then try to update it.
1502 1503 1504 1505 1506 1507
 *
 * 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().
1508 1509 1510
 */
int nfs_post_op_update_inode(struct inode *inode, struct nfs_fattr *fattr)
{
1511
	int status;
1512

1513
	spin_lock(&inode->i_lock);
1514
	nfs_fattr_set_barrier(fattr);
1515
	status = nfs_post_op_update_inode_locked(inode, fattr);
1516
	spin_unlock(&inode->i_lock);
1517

1518
	return status;
1519
}
B
Bryan Schumaker 已提交
1520
EXPORT_SYMBOL_GPL(nfs_post_op_update_inode);
1521

1522
/**
1523
 * nfs_post_op_update_inode_force_wcc_locked - update the inode attribute cache
1524 1525 1526 1527 1528 1529 1530 1531 1532
 * @inode - pointer to inode
 * @fattr - updated attributes
 *
 * 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.
 */
1533
int nfs_post_op_update_inode_force_wcc_locked(struct inode *inode, struct nfs_fattr *fattr)
1534
{
1535 1536 1537 1538 1539
	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)) {
1540 1541 1542 1543
		fattr->valid &= ~(NFS_ATTR_FATTR_PRECHANGE
				| NFS_ATTR_FATTR_PRESIZE
				| NFS_ATTR_FATTR_PREMTIME
				| NFS_ATTR_FATTR_PRECTIME);
1544 1545
		goto out_noforce;
	}
1546 1547
	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRECHANGE) == 0) {
1548
		fattr->pre_change_attr = inode->i_version;
1549
		fattr->valid |= NFS_ATTR_FATTR_PRECHANGE;
1550
	}
1551 1552
	if ((fattr->valid & NFS_ATTR_FATTR_CTIME) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRECTIME) == 0) {
1553
		memcpy(&fattr->pre_ctime, &inode->i_ctime, sizeof(fattr->pre_ctime));
1554 1555 1556 1557
		fattr->valid |= NFS_ATTR_FATTR_PRECTIME;
	}
	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PREMTIME) == 0) {
1558
		memcpy(&fattr->pre_mtime, &inode->i_mtime, sizeof(fattr->pre_mtime));
1559 1560 1561 1562
		fattr->valid |= NFS_ATTR_FATTR_PREMTIME;
	}
	if ((fattr->valid & NFS_ATTR_FATTR_SIZE) != 0 &&
			(fattr->valid & NFS_ATTR_FATTR_PRESIZE) == 0) {
1563
		fattr->pre_size = i_size_read(inode);
1564
		fattr->valid |= NFS_ATTR_FATTR_PRESIZE;
1565
	}
1566 1567
out_noforce:
	status = nfs_post_op_update_inode_locked(inode, fattr);
1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
	return status;
}

/**
 * nfs_post_op_update_inode_force_wcc - try to update the inode attribute cache
 * @inode - pointer to inode
 * @fattr - updated attributes
 *
 * 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);
1587
	nfs_fattr_set_barrier(fattr);
1588
	status = nfs_post_op_update_inode_force_wcc_locked(inode, fattr);
1589 1590
	spin_unlock(&inode->i_lock);
	return status;
1591
}
B
Bryan Schumaker 已提交
1592
EXPORT_SYMBOL_GPL(nfs_post_op_update_inode_force_wcc);
1593

1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606

static inline bool nfs_fileid_valid(struct nfs_inode *nfsi,
				    struct nfs_fattr *fattr)
{
	bool ret1 = true, ret2 = true;

	if (fattr->valid & NFS_ATTR_FATTR_FILEID)
		ret1 = (nfsi->fileid == fattr->fileid);
	if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID)
		ret2 = (nfsi->fileid == fattr->mounted_on_fileid);
	return ret1 || ret2;
}

L
Linus Torvalds 已提交
1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
/*
 * 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.
 */
1619
static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
L
Linus Torvalds 已提交
1620
{
1621
	struct nfs_server *server;
L
Linus Torvalds 已提交
1622
	struct nfs_inode *nfsi = NFS_I(inode);
1623
	loff_t cur_isize, new_isize;
1624
	unsigned long invalid = 0;
1625
	unsigned long now = jiffies;
1626
	unsigned long save_cache_validity;
L
Linus Torvalds 已提交
1627

1628
	dfprintk(VFS, "NFS: %s(%s/%lu fh_crc=0x%08x ct=%d info=0x%x)\n",
1629
			__func__, inode->i_sb->s_id, inode->i_ino,
1630
			nfs_display_fhandle_hash(NFS_FH(inode)),
L
Linus Torvalds 已提交
1631 1632
			atomic_read(&inode->i_count), fattr->valid);

1633
	if (!nfs_fileid_valid(nfsi, fattr)) {
1634 1635 1636 1637 1638 1639 1640
		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
1641 1642 1643 1644

	/*
	 * Make sure the inode's type hasn't changed.
	 */
1645 1646 1647 1648
	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.
		*/
1649
		printk(KERN_DEBUG "NFS: %s: inode %lu mode changed, %07o to %07o\n",
1650 1651 1652
				__func__, inode->i_ino, inode->i_mode, fattr->mode);
		goto out_err;
	}
L
Linus Torvalds 已提交
1653

1654
	server = NFS_SERVER(inode);
T
Trond Myklebust 已提交
1655
	/* Update the fsid? */
1656
	if (S_ISDIR(inode->i_mode) && (fattr->valid & NFS_ATTR_FATTR_FSID) &&
1657
			!nfs_fsid_equal(&server->fsid, &fattr->fsid) &&
1658
			!IS_AUTOMOUNT(inode))
1659 1660
		server->fsid = fattr->fsid;

L
Linus Torvalds 已提交
1661 1662 1663
	/*
	 * Update the read time so we don't revalidate too often.
	 */
1664
	nfsi->read_cache_jiffies = fattr->time_start;
1665

1666 1667 1668 1669 1670
	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
1671

1672
	/* Do atomic weak cache consistency updates */
1673
	invalid |= nfs_wcc_update_inode(inode, fattr);
1674

1675
	/* More cache consistency checks */
1676
	if (fattr->valid & NFS_ATTR_FATTR_CHANGE) {
1677
		if (inode->i_version != fattr->change_attr) {
1678 1679
			dprintk("NFS: change_attr change on server for file %s/%ld\n",
					inode->i_sb->s_id, inode->i_ino);
1680 1681 1682
			invalid |= NFS_INO_INVALID_ATTR
				| NFS_INO_INVALID_DATA
				| NFS_INO_INVALID_ACCESS
1683
				| NFS_INO_INVALID_ACL;
1684 1685
			if (S_ISDIR(inode->i_mode))
				nfs_force_lookup_revalidate(inode);
1686
			inode->i_version = fattr->change_attr;
1687
		}
1688
	} else
1689
		nfsi->cache_validity |= save_cache_validity;
1690 1691

	if (fattr->valid & NFS_ATTR_FATTR_MTIME) {
1692
		memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime));
1693
	} else if (server->caps & NFS_CAP_MTIME)
1694 1695
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1696 1697
				| NFS_INO_REVAL_FORCED);

1698
	if (fattr->valid & NFS_ATTR_FATTR_CTIME) {
1699
		memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime));
1700
	} else if (server->caps & NFS_CAP_CTIME)
1701 1702
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1703
				| NFS_INO_REVAL_FORCED);
1704

1705
	/* Check if our cached file size is stale */
1706 1707 1708 1709 1710 1711
	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
		new_isize = nfs_size_to_loff_t(fattr->size);
		cur_isize = i_size_read(inode);
		if (new_isize != cur_isize) {
			/* Do we perhaps have any outstanding writes, or has
			 * the file grown beyond our last write? */
1712
			if ((nfsi->nrequests == 0) || new_isize > cur_isize) {
1713 1714 1715
				i_size_write(inode, new_isize);
				invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
			}
1716 1717 1718 1719 1720 1721
			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
1722
		}
1723
	} else
1724 1725
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1726 1727
				| NFS_INO_REVAL_PAGECACHE
				| NFS_INO_REVAL_FORCED);
L
Linus Torvalds 已提交
1728 1729


1730 1731
	if (fattr->valid & NFS_ATTR_FATTR_ATIME)
		memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
1732
	else if (server->caps & NFS_CAP_ATIME)
1733 1734
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATIME
1735
				| NFS_INO_REVAL_FORCED);
L
Linus Torvalds 已提交
1736

1737 1738
	if (fattr->valid & NFS_ATTR_FATTR_MODE) {
		if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) {
1739 1740 1741
			umode_t newmode = inode->i_mode & S_IFMT;
			newmode |= fattr->mode & S_IALLUGO;
			inode->i_mode = newmode;
1742 1743
			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
		}
1744
	} else if (server->caps & NFS_CAP_MODE)
1745 1746
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1747 1748 1749 1750
				| NFS_INO_INVALID_ACCESS
				| NFS_INO_INVALID_ACL
				| NFS_INO_REVAL_FORCED);

1751
	if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
1752
		if (!uid_eq(inode->i_uid, fattr->uid)) {
1753 1754 1755
			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
			inode->i_uid = fattr->uid;
		}
1756
	} else if (server->caps & NFS_CAP_OWNER)
1757 1758
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1759 1760 1761 1762
				| NFS_INO_INVALID_ACCESS
				| NFS_INO_INVALID_ACL
				| NFS_INO_REVAL_FORCED);

1763
	if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
1764
		if (!gid_eq(inode->i_gid, fattr->gid)) {
1765 1766 1767
			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
			inode->i_gid = fattr->gid;
		}
1768
	} else if (server->caps & NFS_CAP_OWNER_GROUP)
1769 1770
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1771 1772 1773
				| NFS_INO_INVALID_ACCESS
				| NFS_INO_INVALID_ACL
				| NFS_INO_REVAL_FORCED);
1774

1775 1776 1777 1778 1779
	if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
		if (inode->i_nlink != fattr->nlink) {
			invalid |= NFS_INO_INVALID_ATTR;
			if (S_ISDIR(inode->i_mode))
				invalid |= NFS_INO_INVALID_DATA;
M
Miklos Szeredi 已提交
1780
			set_nlink(inode, fattr->nlink);
1781
		}
1782
	} else if (server->caps & NFS_CAP_NLINK)
1783 1784
		nfsi->cache_validity |= save_cache_validity &
				(NFS_INO_INVALID_ATTR
1785
				| NFS_INO_REVAL_FORCED);
L
Linus Torvalds 已提交
1786

1787
	if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
L
Linus Torvalds 已提交
1788 1789 1790 1791 1792
		/*
		 * report the blocks in 512byte units
		 */
		inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
 	}
1793 1794
	if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
		inode->i_blocks = fattr->du.nfs2.blocks;
L
Linus Torvalds 已提交
1795 1796

	/* Update attrtimeo value if we're out of the unstable period */
1797
	if (invalid & NFS_INO_INVALID_ATTR) {
C
Chuck Lever 已提交
1798
		nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE);
L
Linus Torvalds 已提交
1799
		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
1800
		nfsi->attrtimeo_timestamp = now;
1801
		/* Set barrier to be more recent than all outstanding updates */
1802
		nfsi->attr_gencount = nfs_inc_attr_generation_counter();
1803
	} else {
1804
		if (!time_in_range_open(now, nfsi->attrtimeo_timestamp, nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
1805 1806 1807 1808
			if ((nfsi->attrtimeo <<= 1) > NFS_MAXATTRTIMEO(inode))
				nfsi->attrtimeo = NFS_MAXATTRTIMEO(inode);
			nfsi->attrtimeo_timestamp = now;
		}
1809 1810 1811
		/* 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
1812
	}
1813
	invalid &= ~NFS_INO_INVALID_ATTR;
L
Linus Torvalds 已提交
1814 1815 1816 1817
	/* 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;
1818
	if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ) ||
1819
			(save_cache_validity & NFS_INO_REVAL_FORCED))
1820
		nfs_set_cache_invalid(inode, invalid);
D
David Howells 已提交
1821

L
Linus Torvalds 已提交
1822
	return 0;
1823
 out_err:
L
Linus Torvalds 已提交
1824 1825 1826 1827 1828 1829 1830 1831 1832
	/*
	 * 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
1833
struct inode *nfs_alloc_inode(struct super_block *sb)
L
Linus Torvalds 已提交
1834 1835
{
	struct nfs_inode *nfsi;
F
Firo Yang 已提交
1836
	nfsi = kmem_cache_alloc(nfs_inode_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
1837 1838
	if (!nfsi)
		return NULL;
1839 1840
	nfsi->flags = 0UL;
	nfsi->cache_validity = 0UL;
B
Bryan Schumaker 已提交
1841
#if IS_ENABLED(CONFIG_NFS_V4)
1842 1843
	nfsi->nfs4_acl = NULL;
#endif /* CONFIG_NFS_V4 */
L
Linus Torvalds 已提交
1844 1845
	return &nfsi->vfs_inode;
}
B
Bryan Schumaker 已提交
1846
EXPORT_SYMBOL_GPL(nfs_alloc_inode);
L
Linus Torvalds 已提交
1847

N
Nick Piggin 已提交
1848
static void nfs_i_callback(struct rcu_head *head)
L
Linus Torvalds 已提交
1849
{
N
Nick Piggin 已提交
1850
	struct inode *inode = container_of(head, struct inode, i_rcu);
L
Linus Torvalds 已提交
1851 1852 1853
	kmem_cache_free(nfs_inode_cachep, NFS_I(inode));
}

N
Nick Piggin 已提交
1854 1855 1856 1857
void nfs_destroy_inode(struct inode *inode)
{
	call_rcu(&inode->i_rcu, nfs_i_callback);
}
B
Bryan Schumaker 已提交
1858
EXPORT_SYMBOL_GPL(nfs_destroy_inode);
N
Nick Piggin 已提交
1859

A
Andrew Morton 已提交
1860 1861
static inline void nfs4_init_once(struct nfs_inode *nfsi)
{
B
Bryan Schumaker 已提交
1862
#if IS_ENABLED(CONFIG_NFS_V4)
A
Andrew Morton 已提交
1863 1864 1865
	INIT_LIST_HEAD(&nfsi->open_states);
	nfsi->delegation = NULL;
	init_rwsem(&nfsi->rwsem);
1866
	nfsi->layout = NULL;
A
Andrew Morton 已提交
1867 1868
#endif
}
D
David Howells 已提交
1869

1870
static void init_once(void *foo)
L
Linus Torvalds 已提交
1871 1872 1873
{
	struct nfs_inode *nfsi = (struct nfs_inode *) foo;

C
Christoph Lameter 已提交
1874 1875 1876 1877
	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
1878
	INIT_LIST_HEAD(&nfsi->commit_info.list);
1879
	nfsi->nrequests = 0;
F
Fred Isaman 已提交
1880
	nfsi->commit_info.ncommit = 0;
1881
	atomic_set(&nfsi->commit_info.rpcs_out, 0);
T
Trond Myklebust 已提交
1882 1883 1884
	atomic_set(&nfsi->silly_count, 1);
	INIT_HLIST_HEAD(&nfsi->silly_list);
	init_waitqueue_head(&nfsi->waitqueue);
C
Christoph Lameter 已提交
1885
	nfs4_init_once(nfsi);
L
Linus Torvalds 已提交
1886
}
1887

D
David Howells 已提交
1888
static int __init nfs_init_inodecache(void)
L
Linus Torvalds 已提交
1889 1890 1891
{
	nfs_inode_cachep = kmem_cache_create("nfs_inode_cache",
					     sizeof(struct nfs_inode),
1892 1893
					     0, (SLAB_RECLAIM_ACCOUNT|
						SLAB_MEM_SPREAD),
1894
					     init_once);
L
Linus Torvalds 已提交
1895 1896 1897 1898 1899 1900
	if (nfs_inode_cachep == NULL)
		return -ENOMEM;

	return 0;
}

1901
static void nfs_destroy_inodecache(void)
L
Linus Torvalds 已提交
1902
{
1903 1904 1905 1906 1907
	/*
	 * Make sure all delayed rcu free inodes are flushed before we
	 * destroy cache.
	 */
	rcu_barrier();
1908
	kmem_cache_destroy(nfs_inode_cachep);
L
Linus Torvalds 已提交
1909 1910
}

T
Trond Myklebust 已提交
1911
struct workqueue_struct *nfsiod_workqueue;
B
Bryan Schumaker 已提交
1912
EXPORT_SYMBOL_GPL(nfsiod_workqueue);
T
Trond Myklebust 已提交
1913 1914 1915 1916 1917 1918 1919 1920

/*
 * start up the nfsiod workqueue
 */
static int nfsiod_start(void)
{
	struct workqueue_struct *wq;
	dprintk("RPC:       creating workqueue nfsiod\n");
1921
	wq = alloc_workqueue("nfsiod", WQ_MEM_RECLAIM, 0);
T
Trond Myklebust 已提交
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
	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);
}

1942
int nfs_net_id;
1943
EXPORT_SYMBOL_GPL(nfs_net_id);
1944 1945 1946

static int nfs_net_init(struct net *net)
{
1947
	nfs_clients_init(net);
1948
	return nfs_fs_proc_net_init(net);
1949 1950 1951 1952
}

static void nfs_net_exit(struct net *net)
{
1953
	nfs_fs_proc_net_exit(net);
1954
	nfs_cleanup_cb_ident_idr(net);
1955 1956 1957 1958 1959 1960 1961 1962 1963
}

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
1964 1965 1966 1967 1968 1969 1970
/*
 * Initialize NFS
 */
static int __init init_nfs_fs(void)
{
	int err;

1971
	err = register_pernet_subsys(&nfs_net_ops);
1972
	if (err < 0)
B
Bryan Schumaker 已提交
1973
		goto out9;
1974

1975 1976
	err = nfs_fscache_register();
	if (err < 0)
B
Bryan Schumaker 已提交
1977
		goto out8;
1978

T
Trond Myklebust 已提交
1979 1980
	err = nfsiod_start();
	if (err)
B
Bryan Schumaker 已提交
1981
		goto out7;
T
Trond Myklebust 已提交
1982

1983 1984
	err = nfs_fs_proc_init();
	if (err)
B
Bryan Schumaker 已提交
1985
		goto out6;
1986

L
Linus Torvalds 已提交
1987 1988
	err = nfs_init_nfspagecache();
	if (err)
B
Bryan Schumaker 已提交
1989
		goto out5;
L
Linus Torvalds 已提交
1990 1991 1992

	err = nfs_init_inodecache();
	if (err)
B
Bryan Schumaker 已提交
1993
		goto out4;
L
Linus Torvalds 已提交
1994 1995 1996

	err = nfs_init_readpagecache();
	if (err)
B
Bryan Schumaker 已提交
1997
		goto out3;
L
Linus Torvalds 已提交
1998 1999 2000

	err = nfs_init_writepagecache();
	if (err)
B
Bryan Schumaker 已提交
2001
		goto out2;
L
Linus Torvalds 已提交
2002 2003 2004

	err = nfs_init_directcache();
	if (err)
B
Bryan Schumaker 已提交
2005
		goto out1;
L
Linus Torvalds 已提交
2006

2007
	rpc_proc_register(&init_net, &nfs_rpcstat);
2008 2009 2010

	err = register_nfs_fs();
	if (err)
2011 2012
		goto out0;

L
Linus Torvalds 已提交
2013
	return 0;
2014
out0:
2015
	rpc_proc_unregister(&init_net, "nfs");
L
Linus Torvalds 已提交
2016
	nfs_destroy_directcache();
B
Bryan Schumaker 已提交
2017
out1:
2018
	nfs_destroy_writepagecache();
B
Bryan Schumaker 已提交
2019
out2:
2020
	nfs_destroy_readpagecache();
B
Bryan Schumaker 已提交
2021
out3:
2022
	nfs_destroy_inodecache();
B
Bryan Schumaker 已提交
2023
out4:
2024
	nfs_destroy_nfspagecache();
B
Bryan Schumaker 已提交
2025
out5:
2026
	nfs_fs_proc_exit();
B
Bryan Schumaker 已提交
2027
out6:
2028
	nfsiod_stop();
B
Bryan Schumaker 已提交
2029
out7:
2030
	nfs_fscache_unregister();
B
Bryan Schumaker 已提交
2031
out8:
2032
	unregister_pernet_subsys(&nfs_net_ops);
B
Bryan Schumaker 已提交
2033
out9:
L
Linus Torvalds 已提交
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043
	return err;
}

static void __exit exit_nfs_fs(void)
{
	nfs_destroy_directcache();
	nfs_destroy_writepagecache();
	nfs_destroy_readpagecache();
	nfs_destroy_inodecache();
	nfs_destroy_nfspagecache();
2044
	nfs_fscache_unregister();
2045
	unregister_pernet_subsys(&nfs_net_ops);
2046
	rpc_proc_unregister(&init_net, "nfs");
D
David Howells 已提交
2047
	unregister_nfs_fs();
2048
	nfs_fs_proc_exit();
T
Trond Myklebust 已提交
2049
	nfsiod_stop();
L
Linus Torvalds 已提交
2050 2051 2052 2053 2054
}

/* Not quite true; I just maintain it */
MODULE_AUTHOR("Olaf Kirch <okir@monad.swb.de>");
MODULE_LICENSE("GPL");
2055
module_param(enable_ino64, bool, 0644);
L
Linus Torvalds 已提交
2056 2057 2058

module_init(init_nfs_fs)
module_exit(exit_nfs_fs)