dir.c 56.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
/*
 *  linux/fs/nfs/dir.c
 *
 *  Copyright (C) 1992  Rick Sladkey
 *
 *  nfs directory handling functions
 *
 * 10 Apr 1996	Added silly rename for unlink	--okir
 * 28 Sep 1996	Improved directory cache --okir
 * 23 Aug 1997  Claus Heine claus@momo.math.rwth-aachen.de 
 *              Re-implemented silly rename for unlink, newly implemented
 *              silly rename for nfs_rename() following the suggestions
 *              of Olaf Kirch (okir) found in this file.
 *              Following Linus comments on my original hack, this version
 *              depends only on the dcache stuff and doesn't touch the inode
 *              layer (iput() and friends).
 *  6 Jun 1999	Cache readdir lookups in the page cache. -DaveM
 */

#include <linux/time.h>
#include <linux/errno.h>
#include <linux/stat.h>
#include <linux/fcntl.h>
#include <linux/string.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/sunrpc/clnt.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_mount.h>
#include <linux/pagemap.h>
32
#include <linux/pagevec.h>
L
Linus Torvalds 已提交
33
#include <linux/namei.h>
34
#include <linux/mount.h>
A
Alexey Dobriyan 已提交
35
#include <linux/sched.h>
B
Bryan Schumaker 已提交
36
#include <linux/vmalloc.h>
37
#include <linux/kmemleak.h>
L
Linus Torvalds 已提交
38 39

#include "delegation.h"
C
Chuck Lever 已提交
40
#include "iostat.h"
41
#include "internal.h"
42
#include "fscache.h"
L
Linus Torvalds 已提交
43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

/* #define NFS_DEBUG_VERBOSE 1 */

static int nfs_opendir(struct inode *, struct file *);
static int nfs_readdir(struct file *, void *, filldir_t);
static struct dentry *nfs_lookup(struct inode *, struct dentry *, struct nameidata *);
static int nfs_create(struct inode *, struct dentry *, int, struct nameidata *);
static int nfs_mkdir(struct inode *, struct dentry *, int);
static int nfs_rmdir(struct inode *, struct dentry *);
static int nfs_unlink(struct inode *, struct dentry *);
static int nfs_symlink(struct inode *, struct dentry *, const char *);
static int nfs_link(struct dentry *, struct inode *, struct dentry *);
static int nfs_mknod(struct inode *, struct dentry *, int, dev_t);
static int nfs_rename(struct inode *, struct dentry *,
		      struct inode *, struct dentry *);
58
static int nfs_fsync_dir(struct file *, int);
59
static loff_t nfs_llseek_dir(struct file *, loff_t, int);
B
Bryan Schumaker 已提交
60
static int nfs_readdir_clear_array(struct page*, gfp_t);
L
Linus Torvalds 已提交
61

62
const struct file_operations nfs_dir_operations = {
63
	.llseek		= nfs_llseek_dir,
L
Linus Torvalds 已提交
64 65 66 67 68 69 70
	.read		= generic_read_dir,
	.readdir	= nfs_readdir,
	.open		= nfs_opendir,
	.release	= nfs_release,
	.fsync		= nfs_fsync_dir,
};

71
const struct inode_operations nfs_dir_inode_operations = {
L
Linus Torvalds 已提交
72 73 74 75 76 77 78 79 80 81 82 83 84 85
	.create		= nfs_create,
	.lookup		= nfs_lookup,
	.link		= nfs_link,
	.unlink		= nfs_unlink,
	.symlink	= nfs_symlink,
	.mkdir		= nfs_mkdir,
	.rmdir		= nfs_rmdir,
	.mknod		= nfs_mknod,
	.rename		= nfs_rename,
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
};

B
Bryan Schumaker 已提交
86 87 88 89
const struct address_space_operations nfs_dir_addr_space_ops = {
	.releasepage = nfs_readdir_clear_array,
};

90
#ifdef CONFIG_NFS_V3
91
const struct inode_operations nfs3_dir_inode_operations = {
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
	.create		= nfs_create,
	.lookup		= nfs_lookup,
	.link		= nfs_link,
	.unlink		= nfs_unlink,
	.symlink	= nfs_symlink,
	.mkdir		= nfs_mkdir,
	.rmdir		= nfs_rmdir,
	.mknod		= nfs_mknod,
	.rename		= nfs_rename,
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
	.listxattr	= nfs3_listxattr,
	.getxattr	= nfs3_getxattr,
	.setxattr	= nfs3_setxattr,
	.removexattr	= nfs3_removexattr,
};
#endif  /* CONFIG_NFS_V3 */

L
Linus Torvalds 已提交
111 112 113
#ifdef CONFIG_NFS_V4

static struct dentry *nfs_atomic_lookup(struct inode *, struct dentry *, struct nameidata *);
114
static int nfs_open_create(struct inode *dir, struct dentry *dentry, int mode, struct nameidata *nd);
115
const struct inode_operations nfs4_dir_inode_operations = {
116
	.create		= nfs_open_create,
L
Linus Torvalds 已提交
117 118 119 120 121 122 123 124 125 126 127
	.lookup		= nfs_atomic_lookup,
	.link		= nfs_link,
	.unlink		= nfs_unlink,
	.symlink	= nfs_symlink,
	.mkdir		= nfs_mkdir,
	.rmdir		= nfs_rmdir,
	.mknod		= nfs_mknod,
	.rename		= nfs_rename,
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
128 129 130
	.getxattr       = nfs4_getxattr,
	.setxattr       = nfs4_setxattr,
	.listxattr      = nfs4_listxattr,
L
Linus Torvalds 已提交
131 132 133 134 135 136 137 138 139 140
};

#endif /* CONFIG_NFS_V4 */

/*
 * Open file
 */
static int
nfs_opendir(struct inode *inode, struct file *filp)
{
141
	int res;
L
Linus Torvalds 已提交
142

C
Chuck Lever 已提交
143
	dfprintk(FILE, "NFS: open dir(%s/%s)\n",
144 145 146 147
			filp->f_path.dentry->d_parent->d_name.name,
			filp->f_path.dentry->d_name.name);

	nfs_inc_stats(inode, NFSIOS_VFSOPEN);
C
Chuck Lever 已提交
148

L
Linus Torvalds 已提交
149
	/* Call generic open code in order to cache credentials */
150
	res = nfs_open(inode, filp);
151 152 153 154 155 156 157
	if (filp->f_path.dentry == filp->f_path.mnt->mnt_root) {
		/* This is a mountpoint, so d_revalidate will never
		 * have been called, so we need to refresh the
		 * inode (for close-open consistency) ourselves.
		 */
		__nfs_revalidate_inode(NFS_SERVER(inode), inode);
	}
L
Linus Torvalds 已提交
158 159 160
	return res;
}

B
Bryan Schumaker 已提交
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
struct nfs_cache_array_entry {
	u64 cookie;
	u64 ino;
	struct qstr string;
};

struct nfs_cache_array {
	unsigned int size;
	int eof_index;
	u64 last_cookie;
	struct nfs_cache_array_entry array[0];
};

#define MAX_READDIR_ARRAY ((PAGE_SIZE - sizeof(struct nfs_cache_array)) / sizeof(struct nfs_cache_array_entry))

B
Bryan Schumaker 已提交
176
typedef __be32 * (*decode_dirent_t)(struct xdr_stream *, struct nfs_entry *, struct nfs_server *, int);
L
Linus Torvalds 已提交
177 178 179 180
typedef struct {
	struct file	*file;
	struct page	*page;
	unsigned long	page_index;
181 182
	u64		*dir_cookie;
	loff_t		current_index;
L
Linus Torvalds 已提交
183
	decode_dirent_t	decode;
B
Bryan Schumaker 已提交
184

185
	unsigned long	timestamp;
186
	unsigned long	gencount;
B
Bryan Schumaker 已提交
187 188 189
	unsigned int	cache_entry_index;
	unsigned int	plus:1;
	unsigned int	eof:1;
L
Linus Torvalds 已提交
190 191
} nfs_readdir_descriptor_t;

B
Bryan Schumaker 已提交
192 193
/*
 * The caller is responsible for calling nfs_readdir_release_array(page)
L
Linus Torvalds 已提交
194 195
 */
static
B
Bryan Schumaker 已提交
196 197
struct nfs_cache_array *nfs_readdir_get_array(struct page *page)
{
198
	void *ptr;
B
Bryan Schumaker 已提交
199 200
	if (page == NULL)
		return ERR_PTR(-EIO);
201 202 203 204
	ptr = kmap(page);
	if (ptr == NULL)
		return ERR_PTR(-ENOMEM);
	return ptr;
B
Bryan Schumaker 已提交
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
}

static
void nfs_readdir_release_array(struct page *page)
{
	kunmap(page);
}

/*
 * we are freeing strings created by nfs_add_to_readdir_array()
 */
static
int nfs_readdir_clear_array(struct page *page, gfp_t mask)
{
	struct nfs_cache_array *array = nfs_readdir_get_array(page);
	int i;
221 222 223

	if (IS_ERR(array))
		return PTR_ERR(array);
B
Bryan Schumaker 已提交
224 225 226 227 228 229 230 231 232 233 234 235
	for (i = 0; i < array->size; i++)
		kfree(array->array[i].string.name);
	nfs_readdir_release_array(page);
	return 0;
}

/*
 * the caller is responsible for freeing qstr.name
 * when called by nfs_readdir_add_to_array, the strings will be freed in
 * nfs_clear_readdir_array()
 */
static
236
int nfs_readdir_make_qstr(struct qstr *string, const char *name, unsigned int len)
B
Bryan Schumaker 已提交
237 238 239
{
	string->len = len;
	string->name = kmemdup(name, len, GFP_KERNEL);
240 241
	if (string->name == NULL)
		return -ENOMEM;
242 243 244 245 246
	/*
	 * Avoid a kmemleak false positive. The pointer to the name is stored
	 * in a page cache page which kmemleak does not scan.
	 */
	kmemleak_not_leak(string->name);
247 248
	string->hash = full_name_hash(name, len);
	return 0;
B
Bryan Schumaker 已提交
249 250 251 252 253 254
}

static
int nfs_readdir_add_to_array(struct nfs_entry *entry, struct page *page)
{
	struct nfs_cache_array *array = nfs_readdir_get_array(page);
255 256 257
	struct nfs_cache_array_entry *cache_entry;
	int ret;

B
Bryan Schumaker 已提交
258 259
	if (IS_ERR(array))
		return PTR_ERR(array);
260
	ret = -ENOSPC;
261 262
	if (array->size >= MAX_READDIR_ARRAY)
		goto out;
B
Bryan Schumaker 已提交
263

264 265 266 267 268 269
	cache_entry = &array->array[array->size];
	cache_entry->cookie = entry->prev_cookie;
	cache_entry->ino = entry->ino;
	ret = nfs_readdir_make_qstr(&cache_entry->string, entry->name, entry->len);
	if (ret)
		goto out;
B
Bryan Schumaker 已提交
270
	array->last_cookie = entry->cookie;
271
	array->size++;
B
Bryan Schumaker 已提交
272 273
	if (entry->eof == 1)
		array->eof_index = array->size;
274
out:
B
Bryan Schumaker 已提交
275
	nfs_readdir_release_array(page);
276
	return ret;
B
Bryan Schumaker 已提交
277 278 279 280 281 282 283 284 285 286 287
}

static
int nfs_readdir_search_for_pos(struct nfs_cache_array *array, nfs_readdir_descriptor_t *desc)
{
	loff_t diff = desc->file->f_pos - desc->current_index;
	unsigned int index;

	if (diff < 0)
		goto out_eof;
	if (diff >= array->size) {
288
		if (array->eof_index >= 0)
B
Bryan Schumaker 已提交
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
			goto out_eof;
		desc->current_index += array->size;
		return -EAGAIN;
	}

	index = (unsigned int)diff;
	*desc->dir_cookie = array->array[index].cookie;
	desc->cache_entry_index = index;
	return 0;
out_eof:
	desc->eof = 1;
	return -EBADCOOKIE;
}

static
int nfs_readdir_search_for_cookie(struct nfs_cache_array *array, nfs_readdir_descriptor_t *desc)
{
	int i;
	int status = -EAGAIN;

	for (i = 0; i < array->size; i++) {
		if (array->array[i].cookie == *desc->dir_cookie) {
			desc->cache_entry_index = i;
			status = 0;
313
			goto out;
B
Bryan Schumaker 已提交
314 315
		}
	}
316 317 318 319 320
	if (i == array->eof_index) {
		desc->eof = 1;
		status = -EBADCOOKIE;
	}
out:
B
Bryan Schumaker 已提交
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
	return status;
}

static
int nfs_readdir_search_array(nfs_readdir_descriptor_t *desc)
{
	struct nfs_cache_array *array;
	int status = -EBADCOOKIE;

	if (desc->dir_cookie == NULL)
		goto out;

	array = nfs_readdir_get_array(desc->page);
	if (IS_ERR(array)) {
		status = PTR_ERR(array);
		goto out;
	}

	if (*desc->dir_cookie == 0)
		status = nfs_readdir_search_for_pos(array, desc);
	else
		status = nfs_readdir_search_for_cookie(array, desc);

	nfs_readdir_release_array(desc->page);
out:
	return status;
}

/* Fill a page with xdr information before transferring to the cache page */
static
B
Bryan Schumaker 已提交
351
int nfs_readdir_xdr_filler(struct page **pages, nfs_readdir_descriptor_t *desc,
B
Bryan Schumaker 已提交
352
			struct nfs_entry *entry, struct file *file, struct inode *inode)
L
Linus Torvalds 已提交
353 354
{
	struct rpc_cred	*cred = nfs_file_cred(file);
355
	unsigned long	timestamp, gencount;
L
Linus Torvalds 已提交
356 357 358 359
	int		error;

 again:
	timestamp = jiffies;
360
	gencount = nfs_inc_attr_generation_counter();
B
Bryan Schumaker 已提交
361
	error = NFS_PROTO(inode)->readdir(file->f_path.dentry, cred, entry->cookie, pages,
L
Linus Torvalds 已提交
362 363 364 365 366
					  NFS_SERVER(inode)->dtsize, desc->plus);
	if (error < 0) {
		/* We requested READDIRPLUS, but the server doesn't grok it */
		if (error == -ENOTSUPP && desc->plus) {
			NFS_SERVER(inode)->caps &= ~NFS_CAP_READDIRPLUS;
B
Benny Halevy 已提交
367
			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
368 369 370 371 372
			desc->plus = 0;
			goto again;
		}
		goto error;
	}
373
	desc->timestamp = timestamp;
374
	desc->gencount = gencount;
B
Bryan Schumaker 已提交
375 376
error:
	return error;
L
Linus Torvalds 已提交
377 378
}

B
Bryan Schumaker 已提交
379 380
/* Fill in an entry based on the xdr code stored in desc->page */
static
381
int xdr_decode(nfs_readdir_descriptor_t *desc, struct nfs_entry *entry, struct xdr_stream *stream)
L
Linus Torvalds 已提交
382
{
B
Bryan Schumaker 已提交
383
	__be32 *p = desc->decode(stream, entry, NFS_SERVER(desc->file->f_path.dentry->d_inode), desc->plus);
L
Linus Torvalds 已提交
384 385 386
	if (IS_ERR(p))
		return PTR_ERR(p);

B
Bryan Schumaker 已提交
387 388 389
	entry->fattr->time_start = desc->timestamp;
	entry->fattr->gencount = desc->gencount;
	return 0;
L
Linus Torvalds 已提交
390 391
}

B
Bryan Schumaker 已提交
392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
static
int nfs_same_file(struct dentry *dentry, struct nfs_entry *entry)
{
	struct nfs_inode *node;
	if (dentry->d_inode == NULL)
		goto different;
	node = NFS_I(dentry->d_inode);
	if (node->fh.size != entry->fh->size)
		goto different;
	if (strncmp(node->fh.data, entry->fh->data, node->fh.size) != 0)
		goto different;
	return 1;
different:
	return 0;
}

static
void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry)
{
411 412 413 414 415 416
	struct qstr filename = {
		.len = entry->len,
		.name = entry->name,
	};
	struct dentry *dentry;
	struct dentry *alias;
B
Bryan Schumaker 已提交
417 418 419
	struct inode *dir = parent->d_inode;
	struct inode *inode;

420 421 422 423 424 425 426
	if (filename.name[0] == '.') {
		if (filename.len == 1)
			return;
		if (filename.len == 2 && filename.name[1] == '.')
			return;
	}
	filename.hash = full_name_hash(filename.name, filename.len);
B
Bryan Schumaker 已提交
427

428
	dentry = d_lookup(parent, &filename);
B
Bryan Schumaker 已提交
429 430 431 432 433 434 435 436 437 438 439
	if (dentry != NULL) {
		if (nfs_same_file(dentry, entry)) {
			nfs_refresh_inode(dentry->d_inode, entry->fattr);
			goto out;
		} else {
			d_drop(dentry);
			dput(dentry);
		}
	}

	dentry = d_alloc(parent, &filename);
440 441 442
	if (dentry == NULL)
		return;

B
Bryan Schumaker 已提交
443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
	dentry->d_op = NFS_PROTO(dir)->dentry_ops;
	inode = nfs_fhget(dentry->d_sb, entry->fh, entry->fattr);
	if (IS_ERR(inode))
		goto out;

	alias = d_materialise_unique(dentry, inode);
	if (IS_ERR(alias))
		goto out;
	else if (alias) {
		nfs_set_verifier(alias, nfs_save_change_attribute(dir));
		dput(alias);
	} else
		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));

out:
	dput(dentry);
}

B
Bryan Schumaker 已提交
461 462
/* Perform conversion from xdr to cache array */
static
463
int nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *entry,
B
Bryan Schumaker 已提交
464
				void *xdr_page, struct page *page, unsigned int buflen)
L
Linus Torvalds 已提交
465
{
466 467
	struct xdr_stream stream;
	struct xdr_buf buf;
B
Bryan Schumaker 已提交
468
	__be32 *ptr = xdr_page;
B
Bryan Schumaker 已提交
469 470
	int status;
	struct nfs_cache_array *array;
471 472 473 474 475 476 477 478 479 480 481

	buf.head->iov_base = xdr_page;
	buf.head->iov_len = buflen;
	buf.tail->iov_len = 0;
	buf.page_base = 0;
	buf.page_len = 0;
	buf.buflen = buf.head->iov_len;
	buf.len = buf.head->iov_len;

	xdr_init_decode(&stream, &buf, ptr);

B
Bryan Schumaker 已提交
482 483 484

	do {
		status = xdr_decode(desc, entry, &stream);
485 486 487
		if (status != 0) {
			if (status == -EAGAIN)
				status = 0;
B
Bryan Schumaker 已提交
488
			break;
489
		}
B
Bryan Schumaker 已提交
490

B
Bryan Schumaker 已提交
491 492
		if (desc->plus == 1)
			nfs_prime_dcache(desc->file->f_path.dentry, entry);
493 494 495 496

		status = nfs_readdir_add_to_array(entry, page);
		if (status != 0)
			break;
B
Bryan Schumaker 已提交
497 498 499 500
	} while (!entry->eof);

	if (status == -EBADCOOKIE && entry->eof) {
		array = nfs_readdir_get_array(page);
501 502 503 504 505
		if (!IS_ERR(array)) {
			array->eof_index = array->size;
			status = 0;
			nfs_readdir_release_array(page);
		}
L
Linus Torvalds 已提交
506
	}
507
	return status;
B
Bryan Schumaker 已提交
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
}

static
void nfs_readdir_free_pagearray(struct page **pages, unsigned int npages)
{
	unsigned int i;
	for (i = 0; i < npages; i++)
		put_page(pages[i]);
}

static
void nfs_readdir_free_large_page(void *ptr, struct page **pages,
		unsigned int npages)
{
	vm_unmap_ram(ptr, npages);
	nfs_readdir_free_pagearray(pages, npages);
}

/*
 * nfs_readdir_large_page will allocate pages that must be freed with a call
 * to nfs_readdir_free_large_page
 */
static
void *nfs_readdir_large_page(struct page **pages, unsigned int npages)
{
	void *ptr;
	unsigned int i;

	for (i = 0; i < npages; i++) {
		struct page *page = alloc_page(GFP_KERNEL);
		if (page == NULL)
			goto out_freepages;
		pages[i] = page;
	}

543
	ptr = vm_map_ram(pages, npages, 0, PAGE_KERNEL);
B
Bryan Schumaker 已提交
544 545 546 547 548
	if (!IS_ERR_OR_NULL(ptr))
		return ptr;
out_freepages:
	nfs_readdir_free_pagearray(pages, i);
	return NULL;
L
Linus Torvalds 已提交
549 550
}

B
Bryan Schumaker 已提交
551 552
static
int nfs_readdir_xdr_to_array(nfs_readdir_descriptor_t *desc, struct page *page, struct inode *inode)
553
{
B
Bryan Schumaker 已提交
554 555
	struct page *pages[NFS_MAX_READDIR_PAGES];
	void *pages_ptr = NULL;
B
Bryan Schumaker 已提交
556 557 558
	struct nfs_entry entry;
	struct file	*file = desc->file;
	struct nfs_cache_array *array;
559
	int status = -ENOMEM;
B
Bryan Schumaker 已提交
560
	unsigned int array_size = ARRAY_SIZE(pages);
B
Bryan Schumaker 已提交
561 562 563 564 565 566 567 568

	entry.prev_cookie = 0;
	entry.cookie = *desc->dir_cookie;
	entry.eof = 0;
	entry.fh = nfs_alloc_fhandle();
	entry.fattr = nfs_alloc_fattr();
	if (entry.fh == NULL || entry.fattr == NULL)
		goto out;
569

B
Bryan Schumaker 已提交
570
	array = nfs_readdir_get_array(page);
571 572 573 574
	if (IS_ERR(array)) {
		status = PTR_ERR(array);
		goto out;
	}
B
Bryan Schumaker 已提交
575 576
	memset(array, 0, sizeof(struct nfs_cache_array));
	array->eof_index = -1;
577

B
Bryan Schumaker 已提交
578 579
	pages_ptr = nfs_readdir_large_page(pages, array_size);
	if (!pages_ptr)
B
Bryan Schumaker 已提交
580 581
		goto out_release_array;
	do {
582
		unsigned int pglen;
B
Bryan Schumaker 已提交
583
		status = nfs_readdir_xdr_filler(pages, desc, &entry, file, inode);
584

B
Bryan Schumaker 已提交
585
		if (status < 0)
586
			break;
587 588
		pglen = status;
		status = nfs_readdir_page_filler(desc, &entry, pages_ptr, page, pglen);
589 590 591 592 593 594
		if (status < 0) {
			if (status == -ENOSPC)
				status = 0;
			break;
		}
	} while (array->eof_index < 0);
B
Bryan Schumaker 已提交
595

B
Bryan Schumaker 已提交
596
	nfs_readdir_free_large_page(pages_ptr, pages, array_size);
B
Bryan Schumaker 已提交
597 598 599 600 601
out_release_array:
	nfs_readdir_release_array(page);
out:
	nfs_free_fattr(entry.fattr);
	nfs_free_fhandle(entry.fh);
602 603 604 605
	return status;
}

/*
B
Bryan Schumaker 已提交
606 607 608 609
 * Now we cache directories properly, by converting xdr information
 * to an array that can be used for lookups later.  This results in
 * fewer cache pages, since we can store more information on each page.
 * We only need to convert from xdr once so future lookups are much simpler
L
Linus Torvalds 已提交
610
 */
B
Bryan Schumaker 已提交
611 612
static
int nfs_readdir_filler(nfs_readdir_descriptor_t *desc, struct page* page)
L
Linus Torvalds 已提交
613
{
614
	struct inode	*inode = desc->file->f_path.dentry->d_inode;
615
	int ret;
L
Linus Torvalds 已提交
616

617 618
	ret = nfs_readdir_xdr_to_array(desc, page, inode);
	if (ret < 0)
B
Bryan Schumaker 已提交
619 620
		goto error;
	SetPageUptodate(page);
L
Linus Torvalds 已提交
621

B
Bryan Schumaker 已提交
622 623 624
	if (invalidate_inode_pages2_range(inode->i_mapping, page->index + 1, -1) < 0) {
		/* Should never happen */
		nfs_zap_mapping(inode, inode->i_mapping);
L
Linus Torvalds 已提交
625
	}
B
Bryan Schumaker 已提交
626 627 628 629
	unlock_page(page);
	return 0;
 error:
	unlock_page(page);
630
	return ret;
B
Bryan Schumaker 已提交
631
}
L
Linus Torvalds 已提交
632

B
Bryan Schumaker 已提交
633 634 635 636 637 638 639 640 641 642
static
void cache_page_release(nfs_readdir_descriptor_t *desc)
{
	page_cache_release(desc->page);
	desc->page = NULL;
}

static
struct page *get_cache_page(nfs_readdir_descriptor_t *desc)
{
643
	return read_cache_page(desc->file->f_path.dentry->d_inode->i_mapping,
B
Bryan Schumaker 已提交
644
			desc->page_index, (filler_t *)nfs_readdir_filler, desc);
L
Linus Torvalds 已提交
645 646 647
}

/*
B
Bryan Schumaker 已提交
648
 * Returns 0 if desc->dir_cookie was found on page desc->page_index
L
Linus Torvalds 已提交
649
 */
B
Bryan Schumaker 已提交
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
static
int find_cache_page(nfs_readdir_descriptor_t *desc)
{
	int res;

	desc->page = get_cache_page(desc);
	if (IS_ERR(desc->page))
		return PTR_ERR(desc->page);

	res = nfs_readdir_search_array(desc);
	if (res == 0)
		return 0;
	cache_page_release(desc);
	return res;
}

/* Search for desc->dir_cookie from the beginning of the page cache */
L
Linus Torvalds 已提交
667 668 669
static inline
int readdir_search_pagecache(nfs_readdir_descriptor_t *desc)
{
670
	int res;
B
Bryan Schumaker 已提交
671

672 673
	if (desc->page_index == 0)
		desc->current_index = 0;
B
Bryan Schumaker 已提交
674 675
	while (1) {
		res = find_cache_page(desc);
L
Linus Torvalds 已提交
676 677
		if (res != -EAGAIN)
			break;
B
Bryan Schumaker 已提交
678
		desc->page_index++;
L
Linus Torvalds 已提交
679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
	}
	return res;
}

static inline unsigned int dt_type(struct inode *inode)
{
	return (inode->i_mode >> 12) & 15;
}

/*
 * Once we've found the start of the dirent within a page: fill 'er up...
 */
static 
int nfs_do_filldir(nfs_readdir_descriptor_t *desc, void *dirent,
		   filldir_t filldir)
{
	struct file	*file = desc->file;
B
Bryan Schumaker 已提交
696 697 698 699 700 701 702
	int i = 0;
	int res = 0;
	struct nfs_cache_array *array = NULL;
	unsigned int d_type = DT_UNKNOWN;
	struct dentry *dentry = NULL;

	array = nfs_readdir_get_array(desc->page);
703 704
	if (IS_ERR(array))
		return PTR_ERR(array);
B
Bryan Schumaker 已提交
705 706 707

	for (i = desc->cache_entry_index; i < array->size; i++) {
		d_type = DT_UNKNOWN;
L
Linus Torvalds 已提交
708

B
Bryan Schumaker 已提交
709 710 711
		res = filldir(dirent, array->array[i].string.name,
			array->array[i].string.len, file->f_pos,
			nfs_compat_user_ino64(array->array[i].ino), d_type);
L
Linus Torvalds 已提交
712 713
		if (res < 0)
			break;
714
		file->f_pos++;
B
Bryan Schumaker 已提交
715 716 717 718 719
		desc->cache_entry_index = i;
		if (i < (array->size-1))
			*desc->dir_cookie = array->array[i+1].cookie;
		else
			*desc->dir_cookie = array->last_cookie;
L
Linus Torvalds 已提交
720
	}
721 722
	if (i == array->eof_index)
		desc->eof = 1;
B
Bryan Schumaker 已提交
723 724 725

	nfs_readdir_release_array(desc->page);
	cache_page_release(desc);
L
Linus Torvalds 已提交
726 727
	if (dentry != NULL)
		dput(dentry);
C
Chuck Lever 已提交
728 729
	dfprintk(DIRCACHE, "NFS: nfs_do_filldir() filling ended @ cookie %Lu; returning = %d\n",
			(unsigned long long)*desc->dir_cookie, res);
L
Linus Torvalds 已提交
730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
	return res;
}

/*
 * If we cannot find a cookie in our cache, we suspect that this is
 * because it points to a deleted file, so we ask the server to return
 * whatever it thinks is the next entry. We then feed this to filldir.
 * If all goes well, we should then be able to find our way round the
 * cache on the next call to readdir_search_pagecache();
 *
 * NOTE: we cannot add the anonymous page to the pagecache because
 *	 the data it contains might not be page aligned. Besides,
 *	 we should already have a complete representation of the
 *	 directory in the page cache by the time we get here.
 */
static inline
int uncached_readdir(nfs_readdir_descriptor_t *desc, void *dirent,
		     filldir_t filldir)
{
	struct page	*page = NULL;
	int		status;
B
Bryan Schumaker 已提交
751
	struct inode *inode = desc->file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
752

C
Chuck Lever 已提交
753 754
	dfprintk(DIRCACHE, "NFS: uncached_readdir() searching for cookie %Lu\n",
			(unsigned long long)*desc->dir_cookie);
L
Linus Torvalds 已提交
755 756 757 758 759 760

	page = alloc_page(GFP_HIGHUSER);
	if (!page) {
		status = -ENOMEM;
		goto out;
	}
B
Bryan Schumaker 已提交
761 762

	if (nfs_readdir_xdr_to_array(desc, page, inode) == -1) {
L
Linus Torvalds 已提交
763 764
		status = -EIO;
		goto out_release;
B
Bryan Schumaker 已提交
765
	}
L
Linus Torvalds 已提交
766

767
	desc->page_index = 0;
B
Bryan Schumaker 已提交
768
	desc->page = page;
L
Linus Torvalds 已提交
769 770 771
	status = nfs_do_filldir(desc, dirent, filldir);

 out:
C
Chuck Lever 已提交
772
	dfprintk(DIRCACHE, "NFS: %s: returns %d\n",
773
			__func__, status);
L
Linus Torvalds 已提交
774 775
	return status;
 out_release:
B
Bryan Schumaker 已提交
776
	cache_page_release(desc);
L
Linus Torvalds 已提交
777 778 779
	goto out;
}

780 781 782
/* The file offset position represents the dirent entry number.  A
   last cookie cache takes care of the common case of reading the
   whole directory.
L
Linus Torvalds 已提交
783 784 785
 */
static int nfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
786
	struct dentry	*dentry = filp->f_path.dentry;
L
Linus Torvalds 已提交
787 788 789
	struct inode	*inode = dentry->d_inode;
	nfs_readdir_descriptor_t my_desc,
			*desc = &my_desc;
790
	int res = -ENOMEM;
L
Linus Torvalds 已提交
791

C
Chuck Lever 已提交
792
	dfprintk(FILE, "NFS: readdir(%s/%s) starting at cookie %llu\n",
C
Chuck Lever 已提交
793 794
			dentry->d_parent->d_name.name, dentry->d_name.name,
			(long long)filp->f_pos);
C
Chuck Lever 已提交
795 796
	nfs_inc_stats(inode, NFSIOS_VFSGETDENTS);

L
Linus Torvalds 已提交
797
	/*
798
	 * filp->f_pos points to the dirent entry number.
799
	 * *desc->dir_cookie has the cookie for the next entry. We have
800 801
	 * to either find the entry with the appropriate number or
	 * revalidate the cookie.
L
Linus Torvalds 已提交
802 803 804 805
	 */
	memset(desc, 0, sizeof(*desc));

	desc->file = filp;
806
	desc->dir_cookie = &nfs_file_open_context(filp)->dir_cookie;
L
Linus Torvalds 已提交
807 808 809
	desc->decode = NFS_PROTO(inode)->decode_dirent;
	desc->plus = NFS_USE_READDIRPLUS(inode);

T
Trond Myklebust 已提交
810
	nfs_block_sillyrename(dentry);
811
	res = nfs_revalidate_mapping(inode, filp->f_mapping);
T
Trond Myklebust 已提交
812 813 814
	if (res < 0)
		goto out;

B
Bryan Schumaker 已提交
815
	while (desc->eof != 1) {
L
Linus Torvalds 已提交
816
		res = readdir_search_pagecache(desc);
817

L
Linus Torvalds 已提交
818 819
		if (res == -EBADCOOKIE) {
			/* This means either end of directory */
B
Bryan Schumaker 已提交
820
			if (*desc->dir_cookie && desc->eof == 0) {
L
Linus Torvalds 已提交
821 822 823 824 825 826 827 828 829
				/* Or that the server has 'lost' a cookie */
				res = uncached_readdir(desc, dirent, filldir);
				if (res >= 0)
					continue;
			}
			res = 0;
			break;
		}
		if (res == -ETOOSMALL && desc->plus) {
B
Benny Halevy 已提交
830
			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
831
			nfs_zap_caches(inode);
832
			desc->page_index = 0;
L
Linus Torvalds 已提交
833
			desc->plus = 0;
B
Bryan Schumaker 已提交
834
			desc->eof = 0;
L
Linus Torvalds 已提交
835 836 837 838 839 840 841 842 843 844 845
			continue;
		}
		if (res < 0)
			break;

		res = nfs_do_filldir(desc, dirent, filldir);
		if (res < 0) {
			res = 0;
			break;
		}
	}
T
Trond Myklebust 已提交
846
out:
T
Trond Myklebust 已提交
847
	nfs_unblock_sillyrename(dentry);
C
Chuck Lever 已提交
848 849
	if (res > 0)
		res = 0;
850
	dfprintk(FILE, "NFS: readdir(%s/%s) returns %d\n",
C
Chuck Lever 已提交
851 852 853
			dentry->d_parent->d_name.name, dentry->d_name.name,
			res);
	return res;
L
Linus Torvalds 已提交
854 855
}

856
static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int origin)
857
{
858 859 860
	struct dentry *dentry = filp->f_path.dentry;
	struct inode *inode = dentry->d_inode;

C
Chuck Lever 已提交
861
	dfprintk(FILE, "NFS: llseek dir(%s/%s, %lld, %d)\n",
862 863 864 865 866
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			offset, origin);

	mutex_lock(&inode->i_mutex);
867 868 869 870 871 872 873 874 875 876 877 878
	switch (origin) {
		case 1:
			offset += filp->f_pos;
		case 0:
			if (offset >= 0)
				break;
		default:
			offset = -EINVAL;
			goto out;
	}
	if (offset != filp->f_pos) {
		filp->f_pos = offset;
879
		nfs_file_open_context(filp)->dir_cookie = 0;
880 881
	}
out:
882
	mutex_unlock(&inode->i_mutex);
883 884 885
	return offset;
}

L
Linus Torvalds 已提交
886 887 888 889
/*
 * All directory operations under NFS are synchronous, so fsync()
 * is a dummy operation.
 */
890
static int nfs_fsync_dir(struct file *filp, int datasync)
L
Linus Torvalds 已提交
891
{
892 893
	struct dentry *dentry = filp->f_path.dentry;

C
Chuck Lever 已提交
894
	dfprintk(FILE, "NFS: fsync dir(%s/%s) datasync %d\n",
C
Chuck Lever 已提交
895 896 897
			dentry->d_parent->d_name.name, dentry->d_name.name,
			datasync);

898
	nfs_inc_stats(dentry->d_inode, NFSIOS_VFSFSYNC);
L
Linus Torvalds 已提交
899 900 901
	return 0;
}

902 903 904 905 906 907 908 909 910 911 912 913
/**
 * nfs_force_lookup_revalidate - Mark the directory as having changed
 * @dir - pointer to directory inode
 *
 * This forces the revalidation code in nfs_lookup_revalidate() to do a
 * full lookup on all child dentries of 'dir' whenever a change occurs
 * on the server that might have invalidated our dcache.
 *
 * The caller should be holding dir->i_lock
 */
void nfs_force_lookup_revalidate(struct inode *dir)
{
914
	NFS_I(dir)->cache_change_attribute++;
915 916
}

L
Linus Torvalds 已提交
917 918 919 920 921
/*
 * A check for whether or not the parent directory has changed.
 * In the case it has, we assume that the dentries are untrustworthy
 * and may need to be looked up again.
 */
922
static int nfs_check_verifier(struct inode *dir, struct dentry *dentry)
L
Linus Torvalds 已提交
923 924 925
{
	if (IS_ROOT(dentry))
		return 1;
926 927
	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONE)
		return 0;
928 929 930 931 932 933 934 935
	if (!nfs_verify_change_attribute(dir, dentry->d_time))
		return 0;
	/* Revalidate nfsi->cache_change_attribute before we declare a match */
	if (nfs_revalidate_inode(NFS_SERVER(dir), dir) < 0)
		return 0;
	if (!nfs_verify_change_attribute(dir, dentry->d_time))
		return 0;
	return 1;
L
Linus Torvalds 已提交
936 937
}

938 939 940 941 942 943 944 945 946 947 948 949 950 951
/*
 * Return the intent data that applies to this particular path component
 *
 * Note that the current set of intents only apply to the very last
 * component of the path.
 * We check for this using LOOKUP_CONTINUE and LOOKUP_PARENT.
 */
static inline unsigned int nfs_lookup_check_intent(struct nameidata *nd, unsigned int mask)
{
	if (nd->flags & (LOOKUP_CONTINUE|LOOKUP_PARENT))
		return 0;
	return nd->flags & mask;
}

952 953 954 955 956 957 958 959
/*
 * Use intent information to check whether or not we're going to do
 * an O_EXCL create using this path component.
 */
static int nfs_is_exclusive_create(struct inode *dir, struct nameidata *nd)
{
	if (NFS_PROTO(dir)->version == 2)
		return 0;
960
	return nd && nfs_lookup_check_intent(nd, LOOKUP_EXCL);
961 962
}

963 964 965 966 967 968 969 970
/*
 * Inode and filehandle revalidation for lookups.
 *
 * We force revalidation in the cases where the VFS sets LOOKUP_REVAL,
 * or if the intent information indicates that we're about to open this
 * particular file and the "nocto" mount flag is not set.
 *
 */
L
Linus Torvalds 已提交
971 972 973 974 975
static inline
int nfs_lookup_verify_inode(struct inode *inode, struct nameidata *nd)
{
	struct nfs_server *server = NFS_SERVER(inode);

976 977
	if (test_bit(NFS_INO_MOUNTPOINT, &NFS_I(inode)->flags))
		return 0;
L
Linus Torvalds 已提交
978 979
	if (nd != NULL) {
		/* VFS wants an on-the-wire revalidation */
980
		if (nd->flags & LOOKUP_REVAL)
L
Linus Torvalds 已提交
981 982
			goto out_force;
		/* This is an open(2) */
983
		if (nfs_lookup_check_intent(nd, LOOKUP_OPEN) != 0 &&
984 985 986
				!(server->flags & NFS_MOUNT_NOCTO) &&
				(S_ISREG(inode->i_mode) ||
				 S_ISDIR(inode->i_mode)))
L
Linus Torvalds 已提交
987
			goto out_force;
988
		return 0;
L
Linus Torvalds 已提交
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006
	}
	return nfs_revalidate_inode(server, inode);
out_force:
	return __nfs_revalidate_inode(server, inode);
}

/*
 * We judge how long we want to trust negative
 * dentries by looking at the parent inode mtime.
 *
 * If parent mtime has changed, we revalidate, else we wait for a
 * period corresponding to the parent's attribute cache timeout value.
 */
static inline
int nfs_neg_need_reval(struct inode *dir, struct dentry *dentry,
		       struct nameidata *nd)
{
	/* Don't revalidate a negative dentry if we're creating a new file */
1007
	if (nd != NULL && nfs_lookup_check_intent(nd, LOOKUP_CREATE) != 0)
L
Linus Torvalds 已提交
1008
		return 0;
1009 1010
	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONEG)
		return 1;
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
	return !nfs_check_verifier(dir, dentry);
}

/*
 * This is called every time the dcache has a lookup hit,
 * and we should check whether we can really trust that
 * lookup.
 *
 * NOTE! The hit can be a negative hit too, don't assume
 * we have an inode!
 *
 * If the parent directory is seen to have changed, we throw out the
 * cached dentry and do a new lookup.
 */
static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
{
	struct inode *dir;
	struct inode *inode;
	struct dentry *parent;
1030 1031
	struct nfs_fh *fhandle = NULL;
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
1032 1033 1034 1035
	int error;

	parent = dget_parent(dentry);
	dir = parent->d_inode;
C
Chuck Lever 已提交
1036
	nfs_inc_stats(dir, NFSIOS_DENTRYREVALIDATE);
L
Linus Torvalds 已提交
1037 1038 1039 1040 1041 1042 1043 1044 1045
	inode = dentry->d_inode;

	if (!inode) {
		if (nfs_neg_need_reval(dir, dentry, nd))
			goto out_bad;
		goto out_valid;
	}

	if (is_bad_inode(inode)) {
C
Chuck Lever 已提交
1046
		dfprintk(LOOKUPCACHE, "%s: %s/%s has dud inode\n",
1047
				__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1048
				dentry->d_name.name);
L
Linus Torvalds 已提交
1049 1050 1051
		goto out_bad;
	}

1052 1053 1054
	if (nfs_have_delegation(inode, FMODE_READ))
		goto out_set_verifier;

L
Linus Torvalds 已提交
1055
	/* Force a full look up iff the parent directory has changed */
1056
	if (!nfs_is_exclusive_create(dir, nd) && nfs_check_verifier(dir, dentry)) {
L
Linus Torvalds 已提交
1057 1058 1059 1060 1061 1062 1063 1064
		if (nfs_lookup_verify_inode(inode, nd))
			goto out_zap_parent;
		goto out_valid;
	}

	if (NFS_STALE(inode))
		goto out_bad;

1065 1066 1067 1068 1069 1070 1071
	error = -ENOMEM;
	fhandle = nfs_alloc_fhandle();
	fattr = nfs_alloc_fattr();
	if (fhandle == NULL || fattr == NULL)
		goto out_error;

	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
L
Linus Torvalds 已提交
1072 1073
	if (error)
		goto out_bad;
1074
	if (nfs_compare_fh(NFS_FH(inode), fhandle))
L
Linus Torvalds 已提交
1075
		goto out_bad;
1076
	if ((error = nfs_refresh_inode(inode, fattr)) != 0)
L
Linus Torvalds 已提交
1077 1078
		goto out_bad;

1079 1080
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
1081
out_set_verifier:
1082
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1083 1084
 out_valid:
	dput(parent);
C
Chuck Lever 已提交
1085
	dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) is valid\n",
1086
			__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1087
			dentry->d_name.name);
L
Linus Torvalds 已提交
1088 1089 1090 1091
	return 1;
out_zap_parent:
	nfs_zap_caches(dir);
 out_bad:
1092
	nfs_mark_for_revalidate(dir);
L
Linus Torvalds 已提交
1093 1094 1095 1096 1097 1098
	if (inode && S_ISDIR(inode->i_mode)) {
		/* Purge readdir caches. */
		nfs_zap_caches(inode);
		/* If we have submounts, don't unhash ! */
		if (have_submounts(dentry))
			goto out_valid;
1099 1100
		if (dentry->d_flags & DCACHE_DISCONNECTED)
			goto out_valid;
L
Linus Torvalds 已提交
1101 1102 1103
		shrink_dcache_parent(dentry);
	}
	d_drop(dentry);
1104 1105
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
L
Linus Torvalds 已提交
1106
	dput(parent);
C
Chuck Lever 已提交
1107
	dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) is invalid\n",
1108
			__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1109
			dentry->d_name.name);
L
Linus Torvalds 已提交
1110
	return 0;
1111 1112 1113 1114 1115 1116 1117 1118
out_error:
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
	dput(parent);
	dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) lookup returned error %d\n",
			__func__, dentry->d_parent->d_name.name,
			dentry->d_name.name, error);
	return error;
L
Linus Torvalds 已提交
1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129
}

/*
 * This is called from dput() when d_count is going to 0.
 */
static int nfs_dentry_delete(struct dentry *dentry)
{
	dfprintk(VFS, "NFS: dentry_delete(%s/%s, %x)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		dentry->d_flags);

1130 1131 1132 1133
	/* Unhash any dentry with a stale inode */
	if (dentry->d_inode != NULL && NFS_STALE(dentry->d_inode))
		return 1;

L
Linus Torvalds 已提交
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
	if (dentry->d_flags & DCACHE_NFSFS_RENAMED) {
		/* Unhash it, so that ->d_iput() would be called */
		return 1;
	}
	if (!(dentry->d_sb->s_flags & MS_ACTIVE)) {
		/* Unhash it, so that ancestors of killed async unlink
		 * files will be cleaned up during umount */
		return 1;
	}
	return 0;

}

1147 1148 1149 1150 1151 1152 1153 1154
static void nfs_drop_nlink(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	if (inode->i_nlink > 0)
		drop_nlink(inode);
	spin_unlock(&inode->i_lock);
}

L
Linus Torvalds 已提交
1155 1156 1157 1158 1159 1160
/*
 * Called when the dentry loses inode.
 * We use it to clean up silly-renamed files.
 */
static void nfs_dentry_iput(struct dentry *dentry, struct inode *inode)
{
1161 1162 1163 1164
	if (S_ISDIR(inode->i_mode))
		/* drop any readdir cache as it could easily be old */
		NFS_I(inode)->cache_validity |= NFS_INO_INVALID_DATA;

L
Linus Torvalds 已提交
1165
	if (dentry->d_flags & DCACHE_NFSFS_RENAMED) {
1166
		drop_nlink(inode);
1167
		nfs_complete_unlink(dentry, inode);
L
Linus Torvalds 已提交
1168 1169 1170 1171
	}
	iput(inode);
}

A
Al Viro 已提交
1172
const struct dentry_operations nfs_dentry_operations = {
L
Linus Torvalds 已提交
1173 1174 1175 1176 1177 1178 1179 1180
	.d_revalidate	= nfs_lookup_revalidate,
	.d_delete	= nfs_dentry_delete,
	.d_iput		= nfs_dentry_iput,
};

static struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, struct nameidata *nd)
{
	struct dentry *res;
T
Trond Myklebust 已提交
1181
	struct dentry *parent;
L
Linus Torvalds 已提交
1182
	struct inode *inode = NULL;
1183 1184
	struct nfs_fh *fhandle = NULL;
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
1185 1186 1187 1188
	int error;

	dfprintk(VFS, "NFS: lookup(%s/%s)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name);
C
Chuck Lever 已提交
1189
	nfs_inc_stats(dir, NFSIOS_VFSLOOKUP);
L
Linus Torvalds 已提交
1190 1191 1192 1193 1194 1195 1196

	res = ERR_PTR(-ENAMETOOLONG);
	if (dentry->d_name.len > NFS_SERVER(dir)->namelen)
		goto out;

	dentry->d_op = NFS_PROTO(dir)->dentry_ops;

1197 1198 1199 1200 1201 1202 1203
	/*
	 * If we're doing an exclusive create, optimize away the lookup
	 * but don't hash the dentry.
	 */
	if (nfs_is_exclusive_create(dir, nd)) {
		d_instantiate(dentry, NULL);
		res = NULL;
1204
		goto out;
1205
	}
L
Linus Torvalds 已提交
1206

1207 1208 1209 1210 1211 1212
	res = ERR_PTR(-ENOMEM);
	fhandle = nfs_alloc_fhandle();
	fattr = nfs_alloc_fattr();
	if (fhandle == NULL || fattr == NULL)
		goto out;

T
Trond Myklebust 已提交
1213 1214 1215
	parent = dentry->d_parent;
	/* Protect against concurrent sillydeletes */
	nfs_block_sillyrename(parent);
1216
	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
L
Linus Torvalds 已提交
1217 1218 1219 1220
	if (error == -ENOENT)
		goto no_entry;
	if (error < 0) {
		res = ERR_PTR(error);
T
Trond Myklebust 已提交
1221
		goto out_unblock_sillyrename;
L
Linus Torvalds 已提交
1222
	}
1223
	inode = nfs_fhget(dentry->d_sb, fhandle, fattr);
1224 1225
	res = (struct dentry *)inode;
	if (IS_ERR(res))
T
Trond Myklebust 已提交
1226
		goto out_unblock_sillyrename;
1227

L
Linus Torvalds 已提交
1228
no_entry:
1229
	res = d_materialise_unique(dentry, inode);
1230 1231
	if (res != NULL) {
		if (IS_ERR(res))
T
Trond Myklebust 已提交
1232
			goto out_unblock_sillyrename;
L
Linus Torvalds 已提交
1233
		dentry = res;
1234
	}
L
Linus Torvalds 已提交
1235
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
T
Trond Myklebust 已提交
1236 1237
out_unblock_sillyrename:
	nfs_unblock_sillyrename(parent);
L
Linus Torvalds 已提交
1238
out:
1239 1240
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246
	return res;
}

#ifdef CONFIG_NFS_V4
static int nfs_open_revalidate(struct dentry *, struct nameidata *);

A
Al Viro 已提交
1247
const struct dentry_operations nfs4_dentry_operations = {
L
Linus Torvalds 已提交
1248 1249 1250 1251 1252
	.d_revalidate	= nfs_open_revalidate,
	.d_delete	= nfs_dentry_delete,
	.d_iput		= nfs_dentry_iput,
};

1253 1254 1255 1256
/*
 * Use intent information to determine whether we need to substitute
 * the NFSv4-style stateful OPEN for the LOOKUP call
 */
T
Trond Myklebust 已提交
1257
static int is_atomic_open(struct nameidata *nd)
L
Linus Torvalds 已提交
1258
{
1259
	if (nd == NULL || nfs_lookup_check_intent(nd, LOOKUP_OPEN) == 0)
L
Linus Torvalds 已提交
1260 1261 1262 1263 1264
		return 0;
	/* NFS does not (yet) have a stateful open for directories */
	if (nd->flags & LOOKUP_DIRECTORY)
		return 0;
	/* Are we trying to write to a read only partition? */
1265 1266
	if (__mnt_is_readonly(nd->path.mnt) &&
	    (nd->intent.open.flags & (O_CREAT|O_TRUNC|FMODE_WRITE)))
L
Linus Torvalds 已提交
1267 1268 1269 1270
		return 0;
	return 1;
}

1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
static struct nfs_open_context *nameidata_to_nfs_open_context(struct dentry *dentry, struct nameidata *nd)
{
	struct path path = {
		.mnt = nd->path.mnt,
		.dentry = dentry,
	};
	struct nfs_open_context *ctx;
	struct rpc_cred *cred;
	fmode_t fmode = nd->intent.open.flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC);

	cred = rpc_lookup_cred();
	if (IS_ERR(cred))
		return ERR_CAST(cred);
	ctx = alloc_nfs_open_context(&path, cred, fmode);
	put_rpccred(cred);
	if (ctx == NULL)
		return ERR_PTR(-ENOMEM);
	return ctx;
}

static int do_open(struct inode *inode, struct file *filp)
{
	nfs_fscache_set_inode_cookie(inode, filp);
	return 0;
}

static int nfs_intent_set_file(struct nameidata *nd, struct nfs_open_context *ctx)
{
	struct file *filp;
	int ret = 0;

	/* If the open_intent is for execute, we have an extra check to make */
	if (ctx->mode & FMODE_EXEC) {
		ret = nfs_may_open(ctx->path.dentry->d_inode,
				ctx->cred,
				nd->intent.open.flags);
		if (ret < 0)
			goto out;
	}
	filp = lookup_instantiate_filp(nd, ctx->path.dentry, do_open);
	if (IS_ERR(filp))
		ret = PTR_ERR(filp);
	else
		nfs_file_set_open_context(filp, ctx);
out:
	put_nfs_open_context(ctx);
	return ret;
}

L
Linus Torvalds 已提交
1320 1321
static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
{
1322 1323
	struct nfs_open_context *ctx;
	struct iattr attr;
L
Linus Torvalds 已提交
1324
	struct dentry *res = NULL;
1325
	struct inode *inode;
1326
	int open_flags;
1327
	int err;
L
Linus Torvalds 已提交
1328

C
Chuck Lever 已提交
1329 1330 1331
	dfprintk(VFS, "NFS: atomic_lookup(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);

L
Linus Torvalds 已提交
1332
	/* Check that we are indeed trying to open this file */
T
Trond Myklebust 已提交
1333
	if (!is_atomic_open(nd))
L
Linus Torvalds 已提交
1334 1335 1336 1337 1338 1339 1340 1341
		goto no_open;

	if (dentry->d_name.len > NFS_SERVER(dir)->namelen) {
		res = ERR_PTR(-ENAMETOOLONG);
		goto out;
	}
	dentry->d_op = NFS_PROTO(dir)->dentry_ops;

1342 1343
	/* Let vfs_create() deal with O_EXCL. Instantiate, but don't hash
	 * the dentry. */
1344
	if (nd->flags & LOOKUP_EXCL) {
1345
		d_instantiate(dentry, NULL);
1346 1347
		goto out;
	}
L
Linus Torvalds 已提交
1348

1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
	ctx = nameidata_to_nfs_open_context(dentry, nd);
	res = ERR_CAST(ctx);
	if (IS_ERR(ctx))
		goto out;

	open_flags = nd->intent.open.flags;
	if (nd->flags & LOOKUP_CREATE) {
		attr.ia_mode = nd->intent.open.create_mode;
		attr.ia_valid = ATTR_MODE;
		if (!IS_POSIXACL(dir))
			attr.ia_mode &= ~current_umask();
	} else {
1361
		open_flags &= ~(O_EXCL | O_CREAT);
1362 1363 1364
		attr.ia_valid = 0;
	}

L
Linus Torvalds 已提交
1365
	/* Open the file on the server */
1366
	nfs_block_sillyrename(dentry->d_parent);
1367
	inode = NFS_PROTO(dir)->open_context(dir, ctx, open_flags, &attr);
1368 1369
	if (IS_ERR(inode)) {
		nfs_unblock_sillyrename(dentry->d_parent);
1370
		put_nfs_open_context(ctx);
1371
		switch (PTR_ERR(inode)) {
L
Linus Torvalds 已提交
1372 1373
			/* Make a negative dentry */
			case -ENOENT:
1374
				d_add(dentry, NULL);
1375 1376
				res = NULL;
				goto out;
L
Linus Torvalds 已提交
1377
			/* This turned out not to be a regular file */
1378 1379
			case -ENOTDIR:
				goto no_open;
L
Linus Torvalds 已提交
1380 1381 1382
			case -ELOOP:
				if (!(nd->intent.open.flags & O_NOFOLLOW))
					goto no_open;
1383
			/* case -EISDIR: */
L
Linus Torvalds 已提交
1384 1385
			/* case -EINVAL: */
			default:
1386
				res = ERR_CAST(inode);
L
Linus Torvalds 已提交
1387 1388
				goto out;
		}
1389
	}
1390
	res = d_add_unique(dentry, inode);
1391
	nfs_unblock_sillyrename(dentry->d_parent);
1392 1393 1394
	if (res != NULL) {
		dput(ctx->path.dentry);
		ctx->path.dentry = dget(res);
L
Linus Torvalds 已提交
1395
		dentry = res;
1396
	}
1397 1398 1399 1400 1401 1402
	err = nfs_intent_set_file(nd, ctx);
	if (err < 0) {
		if (res != NULL)
			dput(res);
		return ERR_PTR(err);
	}
L
Linus Torvalds 已提交
1403
out:
1404
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1405 1406 1407 1408 1409 1410 1411 1412 1413 1414
	return res;
no_open:
	return nfs_lookup(dir, dentry, nd);
}

static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
{
	struct dentry *parent = NULL;
	struct inode *inode = dentry->d_inode;
	struct inode *dir;
1415
	struct nfs_open_context *ctx;
L
Linus Torvalds 已提交
1416 1417
	int openflags, ret = 0;

1418
	if (!is_atomic_open(nd) || d_mountpoint(dentry))
T
Trond Myklebust 已提交
1419
		goto no_open;
1420

L
Linus Torvalds 已提交
1421 1422
	parent = dget_parent(dentry);
	dir = parent->d_inode;
1423

L
Linus Torvalds 已提交
1424 1425 1426
	/* We can't create new files in nfs_open_revalidate(), so we
	 * optimize away revalidation of negative dentries.
	 */
1427 1428 1429
	if (inode == NULL) {
		if (!nfs_neg_need_reval(dir, dentry, nd))
			ret = 1;
L
Linus Torvalds 已提交
1430
		goto out;
1431 1432
	}

L
Linus Torvalds 已提交
1433 1434
	/* NFS only supports OPEN on regular files */
	if (!S_ISREG(inode->i_mode))
T
Trond Myklebust 已提交
1435
		goto no_open_dput;
L
Linus Torvalds 已提交
1436 1437 1438
	openflags = nd->intent.open.flags;
	/* We cannot do exclusive creation on a positive dentry */
	if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL))
T
Trond Myklebust 已提交
1439
		goto no_open_dput;
L
Linus Torvalds 已提交
1440
	/* We can't create new files, or truncate existing ones here */
1441
	openflags &= ~(O_CREAT|O_EXCL|O_TRUNC);
L
Linus Torvalds 已提交
1442

1443 1444 1445 1446
	ctx = nameidata_to_nfs_open_context(dentry, nd);
	ret = PTR_ERR(ctx);
	if (IS_ERR(ctx))
		goto out;
L
Linus Torvalds 已提交
1447
	/*
1448
	 * Note: we're not holding inode->i_mutex and so may be racing with
L
Linus Torvalds 已提交
1449 1450 1451
	 * operations that change the directory. We therefore save the
	 * change attribute *before* we do the RPC call.
	 */
1452
	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, NULL);
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466
	if (IS_ERR(inode)) {
		ret = PTR_ERR(inode);
		switch (ret) {
		case -EPERM:
		case -EACCES:
		case -EDQUOT:
		case -ENOSPC:
		case -EROFS:
			goto out_put_ctx;
		default:
			goto out_drop;
		}
	}
	iput(inode);
1467
	if (inode != dentry->d_inode)
1468
		goto out_drop;
1469 1470 1471 1472 1473

	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
	ret = nfs_intent_set_file(nd, ctx);
	if (ret >= 0)
		ret = 1;
L
Linus Torvalds 已提交
1474 1475 1476
out:
	dput(parent);
	return ret;
1477 1478 1479 1480 1481 1482 1483
out_drop:
	d_drop(dentry);
	ret = 0;
out_put_ctx:
	put_nfs_open_context(ctx);
	goto out;

T
Trond Myklebust 已提交
1484
no_open_dput:
L
Linus Torvalds 已提交
1485
	dput(parent);
T
Trond Myklebust 已提交
1486
no_open:
L
Linus Torvalds 已提交
1487 1488
	return nfs_lookup_revalidate(dentry, nd);
}
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509

static int nfs_open_create(struct inode *dir, struct dentry *dentry, int mode,
		struct nameidata *nd)
{
	struct nfs_open_context *ctx = NULL;
	struct iattr attr;
	int error;
	int open_flags = 0;

	dfprintk(VFS, "NFS: create(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);

	attr.ia_mode = mode;
	attr.ia_valid = ATTR_MODE;

	if ((nd->flags & LOOKUP_CREATE) != 0) {
		open_flags = nd->intent.open.flags;

		ctx = nameidata_to_nfs_open_context(dentry, nd);
		error = PTR_ERR(ctx);
		if (IS_ERR(ctx))
1510
			goto out_err_drop;
1511 1512 1513 1514 1515
	}

	error = NFS_PROTO(dir)->create(dir, dentry, &attr, open_flags, ctx);
	if (error != 0)
		goto out_put_ctx;
1516 1517 1518 1519 1520
	if (ctx != NULL) {
		error = nfs_intent_set_file(nd, ctx);
		if (error < 0)
			goto out_err;
	}
1521 1522 1523 1524
	return 0;
out_put_ctx:
	if (ctx != NULL)
		put_nfs_open_context(ctx);
1525
out_err_drop:
1526
	d_drop(dentry);
1527
out_err:
1528 1529 1530
	return error;
}

L
Linus Torvalds 已提交
1531 1532 1533 1534 1535 1536 1537 1538
#endif /* CONFIG_NFSV4 */

/*
 * Code common to create, mkdir, and mknod.
 */
int nfs_instantiate(struct dentry *dentry, struct nfs_fh *fhandle,
				struct nfs_fattr *fattr)
{
1539 1540
	struct dentry *parent = dget_parent(dentry);
	struct inode *dir = parent->d_inode;
L
Linus Torvalds 已提交
1541 1542 1543
	struct inode *inode;
	int error = -EACCES;

1544 1545
	d_drop(dentry);

L
Linus Torvalds 已提交
1546 1547
	/* We may have been initialized further down */
	if (dentry->d_inode)
1548
		goto out;
L
Linus Torvalds 已提交
1549 1550 1551
	if (fhandle->size == 0) {
		error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
		if (error)
1552
			goto out_error;
L
Linus Torvalds 已提交
1553
	}
1554
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1555 1556
	if (!(fattr->valid & NFS_ATTR_FATTR)) {
		struct nfs_server *server = NFS_SB(dentry->d_sb);
1557
		error = server->nfs_client->rpc_ops->getattr(server, fhandle, fattr);
L
Linus Torvalds 已提交
1558
		if (error < 0)
1559
			goto out_error;
L
Linus Torvalds 已提交
1560 1561
	}
	inode = nfs_fhget(dentry->d_sb, fhandle, fattr);
1562 1563
	error = PTR_ERR(inode);
	if (IS_ERR(inode))
1564 1565 1566 1567
		goto out_error;
	d_add(dentry, inode);
out:
	dput(parent);
L
Linus Torvalds 已提交
1568
	return 0;
1569 1570 1571 1572
out_error:
	nfs_mark_for_revalidate(dir);
	dput(parent);
	return error;
L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
}

/*
 * Following a failed create operation, we drop the dentry rather
 * than retain a negative dentry. This avoids a problem in the event
 * that the operation succeeded on the server, but an error in the
 * reply path made it appear to have failed.
 */
static int nfs_create(struct inode *dir, struct dentry *dentry, int mode,
		struct nameidata *nd)
{
	struct iattr attr;
	int error;

C
Chuck Lever 已提交
1587 1588
	dfprintk(VFS, "NFS: create(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
L
Linus Torvalds 已提交
1589 1590 1591 1592

	attr.ia_mode = mode;
	attr.ia_valid = ATTR_MODE;

1593
	error = NFS_PROTO(dir)->create(dir, dentry, &attr, 0, NULL);
L
Linus Torvalds 已提交
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610
	if (error != 0)
		goto out_err;
	return 0;
out_err:
	d_drop(dentry);
	return error;
}

/*
 * See comments for nfs_proc_create regarding failed operations.
 */
static int
nfs_mknod(struct inode *dir, struct dentry *dentry, int mode, dev_t rdev)
{
	struct iattr attr;
	int status;

C
Chuck Lever 已提交
1611 1612
	dfprintk(VFS, "NFS: mknod(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
L
Linus Torvalds 已提交
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636

	if (!new_valid_dev(rdev))
		return -EINVAL;

	attr.ia_mode = mode;
	attr.ia_valid = ATTR_MODE;

	status = NFS_PROTO(dir)->mknod(dir, dentry, &attr, rdev);
	if (status != 0)
		goto out_err;
	return 0;
out_err:
	d_drop(dentry);
	return status;
}

/*
 * See comments for nfs_proc_create regarding failed operations.
 */
static int nfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
	struct iattr attr;
	int error;

C
Chuck Lever 已提交
1637 1638
	dfprintk(VFS, "NFS: mkdir(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
L
Linus Torvalds 已提交
1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651

	attr.ia_valid = ATTR_MODE;
	attr.ia_mode = mode | S_IFDIR;

	error = NFS_PROTO(dir)->mkdir(dir, dentry, &attr);
	if (error != 0)
		goto out_err;
	return 0;
out_err:
	d_drop(dentry);
	return error;
}

1652 1653 1654 1655 1656 1657
static void nfs_dentry_handle_enoent(struct dentry *dentry)
{
	if (dentry->d_inode != NULL && !d_unhashed(dentry))
		d_delete(dentry);
}

L
Linus Torvalds 已提交
1658 1659 1660 1661
static int nfs_rmdir(struct inode *dir, struct dentry *dentry)
{
	int error;

C
Chuck Lever 已提交
1662 1663
	dfprintk(VFS, "NFS: rmdir(%s/%ld), %s\n",
			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
L
Linus Torvalds 已提交
1664 1665 1666 1667

	error = NFS_PROTO(dir)->rmdir(dir, &dentry->d_name);
	/* Ensure the VFS deletes this inode */
	if (error == 0 && dentry->d_inode != NULL)
1668
		clear_nlink(dentry->d_inode);
1669 1670
	else if (error == -ENOENT)
		nfs_dentry_handle_enoent(dentry);
L
Linus Torvalds 已提交
1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697

	return error;
}

/*
 * Remove a file after making sure there are no pending writes,
 * and after checking that the file has only one user. 
 *
 * We invalidate the attribute cache and free the inode prior to the operation
 * to avoid possible races if the server reuses the inode.
 */
static int nfs_safe_remove(struct dentry *dentry)
{
	struct inode *dir = dentry->d_parent->d_inode;
	struct inode *inode = dentry->d_inode;
	int error = -EBUSY;
		
	dfprintk(VFS, "NFS: safe_remove(%s/%s)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name);

	/* If the dentry was sillyrenamed, we simply call d_delete() */
	if (dentry->d_flags & DCACHE_NFSFS_RENAMED) {
		error = 0;
		goto out;
	}

	if (inode != NULL) {
1698
		nfs_inode_return_delegation(inode);
L
Linus Torvalds 已提交
1699 1700 1701
		error = NFS_PROTO(dir)->remove(dir, &dentry->d_name);
		/* The VFS may want to delete this inode */
		if (error == 0)
1702
			nfs_drop_nlink(inode);
1703
		nfs_mark_for_revalidate(inode);
L
Linus Torvalds 已提交
1704 1705
	} else
		error = NFS_PROTO(dir)->remove(dir, &dentry->d_name);
1706 1707
	if (error == -ENOENT)
		nfs_dentry_handle_enoent(dentry);
L
Linus Torvalds 已提交
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729
out:
	return error;
}

/*  We do silly rename. In case sillyrename() returns -EBUSY, the inode
 *  belongs to an active ".nfs..." file and we return -EBUSY.
 *
 *  If sillyrename() returns 0, we do nothing, otherwise we unlink.
 */
static int nfs_unlink(struct inode *dir, struct dentry *dentry)
{
	int error;
	int need_rehash = 0;

	dfprintk(VFS, "NFS: unlink(%s/%ld, %s)\n", dir->i_sb->s_id,
		dir->i_ino, dentry->d_name.name);

	spin_lock(&dcache_lock);
	spin_lock(&dentry->d_lock);
	if (atomic_read(&dentry->d_count) > 1) {
		spin_unlock(&dentry->d_lock);
		spin_unlock(&dcache_lock);
T
Trond Myklebust 已提交
1730 1731
		/* Start asynchronous writeout of the inode */
		write_inode_now(dentry->d_inode, 0);
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
		error = nfs_sillyrename(dir, dentry);
		return error;
	}
	if (!d_unhashed(dentry)) {
		__d_drop(dentry);
		need_rehash = 1;
	}
	spin_unlock(&dentry->d_lock);
	spin_unlock(&dcache_lock);
	error = nfs_safe_remove(dentry);
1742
	if (!error || error == -ENOENT) {
L
Linus Torvalds 已提交
1743 1744 1745 1746 1747 1748
		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
	} else if (need_rehash)
		d_rehash(dentry);
	return error;
}

1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764
/*
 * To create a symbolic link, most file systems instantiate a new inode,
 * add a page to it containing the path, then write it out to the disk
 * using prepare_write/commit_write.
 *
 * Unfortunately the NFS client can't create the in-core inode first
 * because it needs a file handle to create an in-core inode (see
 * fs/nfs/inode.c:nfs_fhget).  We only have a file handle *after* the
 * symlink request has completed on the server.
 *
 * So instead we allocate a raw page, copy the symname into it, then do
 * the SYMLINK request with the page as the buffer.  If it succeeds, we
 * now have a new file handle and can instantiate an in-core NFS inode
 * and move the raw page into its mapping.
 */
static int nfs_symlink(struct inode *dir, struct dentry *dentry, const char *symname)
L
Linus Torvalds 已提交
1765
{
1766 1767 1768
	struct pagevec lru_pvec;
	struct page *page;
	char *kaddr;
L
Linus Torvalds 已提交
1769
	struct iattr attr;
1770
	unsigned int pathlen = strlen(symname);
L
Linus Torvalds 已提交
1771 1772 1773 1774 1775
	int error;

	dfprintk(VFS, "NFS: symlink(%s/%ld, %s, %s)\n", dir->i_sb->s_id,
		dir->i_ino, dentry->d_name.name, symname);

1776 1777
	if (pathlen > PAGE_SIZE)
		return -ENAMETOOLONG;
L
Linus Torvalds 已提交
1778

1779 1780
	attr.ia_mode = S_IFLNK | S_IRWXUGO;
	attr.ia_valid = ATTR_MODE;
L
Linus Torvalds 已提交
1781

1782
	page = alloc_page(GFP_HIGHUSER);
1783
	if (!page)
1784 1785 1786 1787 1788 1789 1790 1791
		return -ENOMEM;

	kaddr = kmap_atomic(page, KM_USER0);
	memcpy(kaddr, symname, pathlen);
	if (pathlen < PAGE_SIZE)
		memset(kaddr + pathlen, 0, PAGE_SIZE - pathlen);
	kunmap_atomic(kaddr, KM_USER0);

1792
	error = NFS_PROTO(dir)->symlink(dir, dentry, page, pathlen, &attr);
1793 1794 1795 1796
	if (error != 0) {
		dfprintk(VFS, "NFS: symlink(%s/%ld, %s, %s) error %d\n",
			dir->i_sb->s_id, dir->i_ino,
			dentry->d_name.name, symname, error);
L
Linus Torvalds 已提交
1797
		d_drop(dentry);
1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
		__free_page(page);
		return error;
	}

	/*
	 * No big deal if we can't add this page to the page cache here.
	 * READLINK will get the missing page from the server if needed.
	 */
	pagevec_init(&lru_pvec, 0);
	if (!add_to_page_cache(page, dentry->d_inode->i_mapping, 0,
							GFP_KERNEL)) {
1809
		pagevec_add(&lru_pvec, page);
1810
		pagevec_lru_add_file(&lru_pvec);
1811 1812 1813 1814 1815 1816
		SetPageUptodate(page);
		unlock_page(page);
	} else
		__free_page(page);

	return 0;
L
Linus Torvalds 已提交
1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828
}

static int 
nfs_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
{
	struct inode *inode = old_dentry->d_inode;
	int error;

	dfprintk(VFS, "NFS: link(%s/%s -> %s/%s)\n",
		old_dentry->d_parent->d_name.name, old_dentry->d_name.name,
		dentry->d_parent->d_name.name, dentry->d_name.name);

1829 1830
	nfs_inode_return_delegation(inode);

1831
	d_drop(dentry);
L
Linus Torvalds 已提交
1832
	error = NFS_PROTO(dir)->link(inode, dir, &dentry->d_name);
1833
	if (error == 0) {
A
Al Viro 已提交
1834
		ihold(inode);
1835
		d_add(dentry, inode);
1836
	}
L
Linus Torvalds 已提交
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
	return error;
}

/*
 * RENAME
 * FIXME: Some nfsds, like the Linux user space nfsd, may generate a
 * different file handle for the same inode after a rename (e.g. when
 * moving to a different directory). A fail-safe method to do so would
 * be to look up old_dir/old_name, create a link to new_dir/new_name and
 * rename the old file using the sillyrename stuff. This way, the original
 * file in old_dir will go away when the last process iput()s the inode.
 *
 * FIXED.
 * 
 * It actually works quite well. One needs to have the possibility for
 * at least one ".nfs..." file in each directory the file ever gets
 * moved or linked to which happens automagically with the new
 * implementation that only depends on the dcache stuff instead of
 * using the inode layer
 *
 * Unfortunately, things are a little more complicated than indicated
 * above. For a cross-directory move, we want to make sure we can get
 * rid of the old inode after the operation.  This means there must be
 * no pending writes (if it's a file), and the use count must be 1.
 * If these conditions are met, we can drop the dentries before doing
 * the rename.
 */
static int nfs_rename(struct inode *old_dir, struct dentry *old_dentry,
		      struct inode *new_dir, struct dentry *new_dentry)
{
	struct inode *old_inode = old_dentry->d_inode;
	struct inode *new_inode = new_dentry->d_inode;
	struct dentry *dentry = NULL, *rehash = NULL;
	int error = -EBUSY;

	dfprintk(VFS, "NFS: rename(%s/%s -> %s/%s, ct=%d)\n",
		 old_dentry->d_parent->d_name.name, old_dentry->d_name.name,
		 new_dentry->d_parent->d_name.name, new_dentry->d_name.name,
		 atomic_read(&new_dentry->d_count));

	/*
M
Miklos Szeredi 已提交
1878 1879 1880 1881
	 * For non-directories, check whether the target is busy and if so,
	 * make a copy of the dentry and then do a silly-rename. If the
	 * silly-rename succeeds, the copied dentry is hashed and becomes
	 * the new target.
L
Linus Torvalds 已提交
1882
	 */
1883 1884 1885 1886 1887 1888 1889 1890 1891
	if (new_inode && !S_ISDIR(new_inode->i_mode)) {
		/*
		 * To prevent any new references to the target during the
		 * rename, we unhash the dentry in advance.
		 */
		if (!d_unhashed(new_dentry)) {
			d_drop(new_dentry);
			rehash = new_dentry;
		}
L
Linus Torvalds 已提交
1892

1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903
		if (atomic_read(&new_dentry->d_count) > 2) {
			int err;

			/* copy the target dentry's name */
			dentry = d_alloc(new_dentry->d_parent,
					 &new_dentry->d_name);
			if (!dentry)
				goto out;

			/* silly-rename the existing target ... */
			err = nfs_sillyrename(new_dir, new_dentry);
1904
			if (err)
1905
				goto out;
1906 1907

			new_dentry = dentry;
O
OGAWA Hirofumi 已提交
1908
			rehash = NULL;
1909
			new_inode = NULL;
1910
		}
1911
	}
L
Linus Torvalds 已提交
1912

1913
	nfs_inode_return_delegation(old_inode);
1914
	if (new_inode != NULL)
1915
		nfs_inode_return_delegation(new_inode);
L
Linus Torvalds 已提交
1916 1917 1918

	error = NFS_PROTO(old_dir)->rename(old_dir, &old_dentry->d_name,
					   new_dir, &new_dentry->d_name);
1919
	nfs_mark_for_revalidate(old_inode);
L
Linus Torvalds 已提交
1920 1921 1922 1923
out:
	if (rehash)
		d_rehash(rehash);
	if (!error) {
1924 1925
		if (new_inode != NULL)
			nfs_drop_nlink(new_inode);
1926
		d_move(old_dentry, new_dentry);
1927 1928
		nfs_set_verifier(new_dentry,
					nfs_save_change_attribute(new_dir));
1929 1930
	} else if (error == -ENOENT)
		nfs_dentry_handle_enoent(old_dentry);
L
Linus Torvalds 已提交
1931 1932 1933 1934 1935 1936 1937

	/* new dentry created? */
	if (dentry)
		dput(dentry);
	return error;
}

1938 1939 1940 1941
static DEFINE_SPINLOCK(nfs_access_lru_lock);
static LIST_HEAD(nfs_access_lru_list);
static atomic_long_t nfs_access_nr_entries;

1942 1943 1944 1945
static void nfs_access_free_entry(struct nfs_access_entry *entry)
{
	put_rpccred(entry->cred);
	kfree(entry);
1946 1947 1948
	smp_mb__before_atomic_dec();
	atomic_long_dec(&nfs_access_nr_entries);
	smp_mb__after_atomic_dec();
1949 1950
}

1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961
static void nfs_access_free_list(struct list_head *head)
{
	struct nfs_access_entry *cache;

	while (!list_empty(head)) {
		cache = list_entry(head->next, struct nfs_access_entry, lru);
		list_del(&cache->lru);
		nfs_access_free_entry(cache);
	}
}

1962
int nfs_access_cache_shrinker(struct shrinker *shrink, int nr_to_scan, gfp_t gfp_mask)
1963 1964
{
	LIST_HEAD(head);
1965
	struct nfs_inode *nfsi, *next;
1966 1967
	struct nfs_access_entry *cache;

1968 1969
	if ((gfp_mask & GFP_KERNEL) != GFP_KERNEL)
		return (nr_to_scan == 0) ? 0 : -1;
1970

1971
	spin_lock(&nfs_access_lru_lock);
1972
	list_for_each_entry_safe(nfsi, next, &nfs_access_lru_list, access_cache_inode_lru) {
1973 1974 1975 1976
		struct inode *inode;

		if (nr_to_scan-- == 0)
			break;
1977
		inode = &nfsi->vfs_inode;
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
		spin_lock(&inode->i_lock);
		if (list_empty(&nfsi->access_cache_entry_lru))
			goto remove_lru_entry;
		cache = list_entry(nfsi->access_cache_entry_lru.next,
				struct nfs_access_entry, lru);
		list_move(&cache->lru, &head);
		rb_erase(&cache->rb_node, &nfsi->access_cache);
		if (!list_empty(&nfsi->access_cache_entry_lru))
			list_move_tail(&nfsi->access_cache_inode_lru,
					&nfs_access_lru_list);
		else {
remove_lru_entry:
			list_del_init(&nfsi->access_cache_inode_lru);
1991
			smp_mb__before_clear_bit();
1992
			clear_bit(NFS_INO_ACL_LRU_SET, &nfsi->flags);
1993
			smp_mb__after_clear_bit();
1994
		}
1995
		spin_unlock(&inode->i_lock);
1996 1997
	}
	spin_unlock(&nfs_access_lru_lock);
1998
	nfs_access_free_list(&head);
1999 2000 2001
	return (atomic_long_read(&nfs_access_nr_entries) / 100) * sysctl_vfs_cache_pressure;
}

2002
static void __nfs_access_zap_cache(struct nfs_inode *nfsi, struct list_head *head)
L
Linus Torvalds 已提交
2003
{
2004
	struct rb_root *root_node = &nfsi->access_cache;
2005
	struct rb_node *n;
2006 2007 2008 2009 2010 2011
	struct nfs_access_entry *entry;

	/* Unhook entries from the cache */
	while ((n = rb_first(root_node)) != NULL) {
		entry = rb_entry(n, struct nfs_access_entry, rb_node);
		rb_erase(n, root_node);
2012
		list_move(&entry->lru, head);
2013 2014
	}
	nfsi->cache_validity &= ~NFS_INO_INVALID_ACCESS;
L
Linus Torvalds 已提交
2015 2016
}

2017
void nfs_access_zap_cache(struct inode *inode)
L
Linus Torvalds 已提交
2018
{
2019 2020 2021 2022
	LIST_HEAD(head);

	if (test_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags) == 0)
		return;
2023
	/* Remove from global LRU init */
2024 2025
	spin_lock(&nfs_access_lru_lock);
	if (test_and_clear_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags))
2026 2027
		list_del_init(&NFS_I(inode)->access_cache_inode_lru);

2028
	spin_lock(&inode->i_lock);
2029 2030 2031 2032
	__nfs_access_zap_cache(NFS_I(inode), &head);
	spin_unlock(&inode->i_lock);
	spin_unlock(&nfs_access_lru_lock);
	nfs_access_free_list(&head);
2033
}
L
Linus Torvalds 已提交
2034

2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048
static struct nfs_access_entry *nfs_access_search_rbtree(struct inode *inode, struct rpc_cred *cred)
{
	struct rb_node *n = NFS_I(inode)->access_cache.rb_node;
	struct nfs_access_entry *entry;

	while (n != NULL) {
		entry = rb_entry(n, struct nfs_access_entry, rb_node);

		if (cred < entry->cred)
			n = n->rb_left;
		else if (cred > entry->cred)
			n = n->rb_right;
		else
			return entry;
L
Linus Torvalds 已提交
2049
	}
2050 2051 2052
	return NULL;
}

2053
static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res)
2054 2055 2056 2057 2058
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_access_entry *cache;
	int err = -ENOENT;

2059
	spin_lock(&inode->i_lock);
2060 2061 2062 2063 2064
	if (nfsi->cache_validity & NFS_INO_INVALID_ACCESS)
		goto out_zap;
	cache = nfs_access_search_rbtree(inode, cred);
	if (cache == NULL)
		goto out;
2065
	if (!nfs_have_delegated_attributes(inode) &&
2066
	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
2067 2068 2069 2070
		goto out_stale;
	res->jiffies = cache->jiffies;
	res->cred = cache->cred;
	res->mask = cache->mask;
2071
	list_move_tail(&cache->lru, &nfsi->access_cache_entry_lru);
2072 2073 2074 2075 2076 2077
	err = 0;
out:
	spin_unlock(&inode->i_lock);
	return err;
out_stale:
	rb_erase(&cache->rb_node, &nfsi->access_cache);
2078
	list_del(&cache->lru);
2079 2080 2081 2082
	spin_unlock(&inode->i_lock);
	nfs_access_free_entry(cache);
	return -ENOENT;
out_zap:
2083 2084
	spin_unlock(&inode->i_lock);
	nfs_access_zap_cache(inode);
2085 2086 2087 2088 2089
	return -ENOENT;
}

static void nfs_access_add_rbtree(struct inode *inode, struct nfs_access_entry *set)
{
2090 2091
	struct nfs_inode *nfsi = NFS_I(inode);
	struct rb_root *root_node = &nfsi->access_cache;
2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109
	struct rb_node **p = &root_node->rb_node;
	struct rb_node *parent = NULL;
	struct nfs_access_entry *entry;

	spin_lock(&inode->i_lock);
	while (*p != NULL) {
		parent = *p;
		entry = rb_entry(parent, struct nfs_access_entry, rb_node);

		if (set->cred < entry->cred)
			p = &parent->rb_left;
		else if (set->cred > entry->cred)
			p = &parent->rb_right;
		else
			goto found;
	}
	rb_link_node(&set->rb_node, parent, p);
	rb_insert_color(&set->rb_node, root_node);
2110
	list_add_tail(&set->lru, &nfsi->access_cache_entry_lru);
2111
	spin_unlock(&inode->i_lock);
2112 2113 2114
	return;
found:
	rb_replace_node(parent, &set->rb_node, root_node);
2115 2116
	list_add_tail(&set->lru, &nfsi->access_cache_entry_lru);
	list_del(&entry->lru);
2117 2118 2119 2120
	spin_unlock(&inode->i_lock);
	nfs_access_free_entry(entry);
}

2121
static void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set)
2122 2123 2124 2125 2126
{
	struct nfs_access_entry *cache = kmalloc(sizeof(*cache), GFP_KERNEL);
	if (cache == NULL)
		return;
	RB_CLEAR_NODE(&cache->rb_node);
L
Linus Torvalds 已提交
2127
	cache->jiffies = set->jiffies;
2128
	cache->cred = get_rpccred(set->cred);
L
Linus Torvalds 已提交
2129
	cache->mask = set->mask;
2130 2131

	nfs_access_add_rbtree(inode, cache);
2132 2133 2134 2135 2136 2137 2138

	/* Update accounting */
	smp_mb__before_atomic_inc();
	atomic_long_inc(&nfs_access_nr_entries);
	smp_mb__after_atomic_inc();

	/* Add inode to global LRU list */
2139
	if (!test_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags)) {
2140
		spin_lock(&nfs_access_lru_lock);
2141 2142 2143
		if (!test_and_set_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags))
			list_add_tail(&NFS_I(inode)->access_cache_inode_lru,
					&nfs_access_lru_list);
2144 2145
		spin_unlock(&nfs_access_lru_lock);
	}
L
Linus Torvalds 已提交
2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
}

static int nfs_do_access(struct inode *inode, struct rpc_cred *cred, int mask)
{
	struct nfs_access_entry cache;
	int status;

	status = nfs_access_get_cached(inode, cred, &cache);
	if (status == 0)
		goto out;

	/* Be clever: ask server to check for all possible rights */
	cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ;
	cache.cred = cred;
	cache.jiffies = jiffies;
	status = NFS_PROTO(inode)->access(inode, &cache);
2162 2163 2164 2165 2166 2167
	if (status != 0) {
		if (status == -ESTALE) {
			nfs_zap_caches(inode);
			if (!S_ISDIR(inode->i_mode))
				set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
		}
L
Linus Torvalds 已提交
2168
		return status;
2169
	}
L
Linus Torvalds 已提交
2170 2171
	nfs_access_add_cache(inode, &cache);
out:
2172
	if ((mask & ~cache.mask & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
L
Linus Torvalds 已提交
2173 2174 2175 2176
		return 0;
	return -EACCES;
}

2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194
static int nfs_open_permission_mask(int openflags)
{
	int mask = 0;

	if (openflags & FMODE_READ)
		mask |= MAY_READ;
	if (openflags & FMODE_WRITE)
		mask |= MAY_WRITE;
	if (openflags & FMODE_EXEC)
		mask |= MAY_EXEC;
	return mask;
}

int nfs_may_open(struct inode *inode, struct rpc_cred *cred, int openflags)
{
	return nfs_do_access(inode, cred, nfs_open_permission_mask(openflags));
}

2195
int nfs_permission(struct inode *inode, int mask)
L
Linus Torvalds 已提交
2196 2197 2198 2199
{
	struct rpc_cred *cred;
	int res = 0;

C
Chuck Lever 已提交
2200 2201
	nfs_inc_stats(inode, NFSIOS_VFSACCESS);

2202
	if ((mask & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
L
Linus Torvalds 已提交
2203 2204
		goto out;
	/* Is this sys_access() ? */
E
Eric Paris 已提交
2205
	if (mask & (MAY_ACCESS | MAY_CHDIR))
L
Linus Torvalds 已提交
2206 2207 2208 2209 2210 2211 2212 2213
		goto force_lookup;

	switch (inode->i_mode & S_IFMT) {
		case S_IFLNK:
			goto out;
		case S_IFREG:
			/* NFSv4 has atomic_open... */
			if (nfs_server_capable(inode, NFS_CAP_ATOMIC_OPEN)
2214 2215
					&& (mask & MAY_OPEN)
					&& !(mask & MAY_EXEC))
L
Linus Torvalds 已提交
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230
				goto out;
			break;
		case S_IFDIR:
			/*
			 * Optimize away all write operations, since the server
			 * will check permissions when we perform the op.
			 */
			if ((mask & MAY_WRITE) && !(mask & MAY_READ))
				goto out;
	}

force_lookup:
	if (!NFS_PROTO(inode)->access)
		goto out_notsup;

2231
	cred = rpc_lookup_cred();
L
Linus Torvalds 已提交
2232 2233 2234 2235 2236 2237
	if (!IS_ERR(cred)) {
		res = nfs_do_access(inode, cred, mask);
		put_rpccred(cred);
	} else
		res = PTR_ERR(cred);
out:
2238 2239 2240
	if (!res && (mask & MAY_EXEC) && !execute_ok(inode))
		res = -EACCES;

C
Chuck Lever 已提交
2241 2242
	dfprintk(VFS, "NFS: permission(%s/%ld), mask=0x%x, res=%d\n",
		inode->i_sb->s_id, inode->i_ino, mask, res);
L
Linus Torvalds 已提交
2243 2244 2245 2246 2247
	return res;
out_notsup:
	res = nfs_revalidate_inode(NFS_SERVER(inode), inode);
	if (res == 0)
		res = generic_permission(inode, mask, NULL);
C
Chuck Lever 已提交
2248
	goto out;
L
Linus Torvalds 已提交
2249 2250 2251 2252 2253 2254 2255 2256
}

/*
 * Local variables:
 *  version-control: t
 *  kept-new-versions: 5
 * End:
 */