dir.c 56.3 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>
L
Linus Torvalds 已提交
37 38

#include "delegation.h"
C
Chuck Lever 已提交
39
#include "iostat.h"
40
#include "internal.h"
41
#include "fscache.h"
L
Linus Torvalds 已提交
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

/* #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 *);
57
static int nfs_fsync_dir(struct file *, int);
58
static loff_t nfs_llseek_dir(struct file *, loff_t, int);
B
Bryan Schumaker 已提交
59
static int nfs_readdir_clear_array(struct page*, gfp_t);
L
Linus Torvalds 已提交
60

61
const struct file_operations nfs_dir_operations = {
62
	.llseek		= nfs_llseek_dir,
L
Linus Torvalds 已提交
63 64 65 66 67 68 69
	.read		= generic_read_dir,
	.readdir	= nfs_readdir,
	.open		= nfs_opendir,
	.release	= nfs_release,
	.fsync		= nfs_fsync_dir,
};

70
const struct inode_operations nfs_dir_inode_operations = {
L
Linus Torvalds 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83 84
	.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
85 86 87 88
const struct address_space_operations nfs_dir_addr_space_ops = {
	.releasepage = nfs_readdir_clear_array,
};

89
#ifdef CONFIG_NFS_V3
90
const struct inode_operations nfs3_dir_inode_operations = {
91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
	.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
110 111 112
#ifdef CONFIG_NFS_V4

static struct dentry *nfs_atomic_lookup(struct inode *, struct dentry *, struct nameidata *);
113
static int nfs_open_create(struct inode *dir, struct dentry *dentry, int mode, struct nameidata *nd);
114
const struct inode_operations nfs4_dir_inode_operations = {
115
	.create		= nfs_open_create,
L
Linus Torvalds 已提交
116 117 118 119 120 121 122 123 124 125 126
	.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,
127 128 129
	.getxattr       = nfs4_getxattr,
	.setxattr       = nfs4_setxattr,
	.listxattr      = nfs4_listxattr,
L
Linus Torvalds 已提交
130 131 132 133 134 135 136 137 138 139
};

#endif /* CONFIG_NFS_V4 */

/*
 * Open file
 */
static int
nfs_opendir(struct inode *inode, struct file *filp)
{
140
	int res;
L
Linus Torvalds 已提交
141

C
Chuck Lever 已提交
142
	dfprintk(FILE, "NFS: open dir(%s/%s)\n",
143 144 145 146
			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
147

L
Linus Torvalds 已提交
148
	/* Call generic open code in order to cache credentials */
149
	res = nfs_open(inode, filp);
150 151 152 153 154 155 156
	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
157 158 159
	return res;
}

B
Bryan Schumaker 已提交
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
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
175
typedef __be32 * (*decode_dirent_t)(struct xdr_stream *, struct nfs_entry *, struct nfs_server *, int);
L
Linus Torvalds 已提交
176 177 178 179
typedef struct {
	struct file	*file;
	struct page	*page;
	unsigned long	page_index;
180 181
	u64		*dir_cookie;
	loff_t		current_index;
L
Linus Torvalds 已提交
182
	decode_dirent_t	decode;
B
Bryan Schumaker 已提交
183

184
	unsigned long	timestamp;
185
	unsigned long	gencount;
B
Bryan Schumaker 已提交
186 187 188
	unsigned int	cache_entry_index;
	unsigned int	plus:1;
	unsigned int	eof:1;
L
Linus Torvalds 已提交
189 190
} nfs_readdir_descriptor_t;

B
Bryan Schumaker 已提交
191 192
/*
 * The caller is responsible for calling nfs_readdir_release_array(page)
L
Linus Torvalds 已提交
193 194
 */
static
B
Bryan Schumaker 已提交
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
struct nfs_cache_array *nfs_readdir_get_array(struct page *page)
{
	if (page == NULL)
		return ERR_PTR(-EIO);
	return (struct nfs_cache_array *)kmap(page);
}

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;
	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
228
int nfs_readdir_make_qstr(struct qstr *string, const char *name, unsigned int len)
B
Bryan Schumaker 已提交
229 230 231
{
	string->len = len;
	string->name = kmemdup(name, len, GFP_KERNEL);
232 233 234 235
	if (string->name == NULL)
		return -ENOMEM;
	string->hash = full_name_hash(name, len);
	return 0;
B
Bryan Schumaker 已提交
236 237 238 239 240 241
}

static
int nfs_readdir_add_to_array(struct nfs_entry *entry, struct page *page)
{
	struct nfs_cache_array *array = nfs_readdir_get_array(page);
242 243 244
	struct nfs_cache_array_entry *cache_entry;
	int ret;

B
Bryan Schumaker 已提交
245 246
	if (IS_ERR(array))
		return PTR_ERR(array);
247 248 249
	ret = -EIO;
	if (array->size >= MAX_READDIR_ARRAY)
		goto out;
B
Bryan Schumaker 已提交
250

251 252 253 254 255 256
	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
257 258 259 260
	array->last_cookie = entry->cookie;
	if (entry->eof == 1)
		array->eof_index = array->size;
	array->size++;
261
out:
B
Bryan Schumaker 已提交
262
	nfs_readdir_release_array(page);
263
	return ret;
B
Bryan Schumaker 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
}

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) {
		if (array->eof_index > 0)
			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;
	if (index == array->eof_index)
		desc->eof = 1;
	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 (i == array->eof_index) {
			desc->eof = 1;
			status = -EBADCOOKIE;
		}
		if (array->array[i].cookie == *desc->dir_cookie) {
			desc->cache_entry_index = i;
			status = 0;
			break;
		}
	}

	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
340
int nfs_readdir_xdr_filler(struct page **pages, nfs_readdir_descriptor_t *desc,
B
Bryan Schumaker 已提交
341
			struct nfs_entry *entry, struct file *file, struct inode *inode)
L
Linus Torvalds 已提交
342 343
{
	struct rpc_cred	*cred = nfs_file_cred(file);
344
	unsigned long	timestamp, gencount;
L
Linus Torvalds 已提交
345 346 347 348
	int		error;

 again:
	timestamp = jiffies;
349
	gencount = nfs_inc_attr_generation_counter();
B
Bryan Schumaker 已提交
350
	error = NFS_PROTO(inode)->readdir(file->f_path.dentry, cred, entry->cookie, pages,
L
Linus Torvalds 已提交
351 352 353 354 355
					  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
356
			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
357 358 359 360 361
			desc->plus = 0;
			goto again;
		}
		goto error;
	}
362
	desc->timestamp = timestamp;
363
	desc->gencount = gencount;
B
Bryan Schumaker 已提交
364 365
error:
	return error;
L
Linus Torvalds 已提交
366 367
}

B
Bryan Schumaker 已提交
368 369
/* Fill in an entry based on the xdr code stored in desc->page */
static
370
int xdr_decode(nfs_readdir_descriptor_t *desc, struct nfs_entry *entry, struct xdr_stream *stream)
L
Linus Torvalds 已提交
371
{
B
Bryan Schumaker 已提交
372
	__be32 *p = desc->decode(stream, entry, NFS_SERVER(desc->file->f_path.dentry->d_inode), desc->plus);
L
Linus Torvalds 已提交
373 374 375
	if (IS_ERR(p))
		return PTR_ERR(p);

B
Bryan Schumaker 已提交
376 377 378
	entry->fattr->time_start = desc->timestamp;
	entry->fattr->gencount = desc->gencount;
	return 0;
L
Linus Torvalds 已提交
379 380
}

B
Bryan Schumaker 已提交
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
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)
{
400 401 402 403 404 405
	struct qstr filename = {
		.len = entry->len,
		.name = entry->name,
	};
	struct dentry *dentry;
	struct dentry *alias;
B
Bryan Schumaker 已提交
406 407 408
	struct inode *dir = parent->d_inode;
	struct inode *inode;

409 410 411 412 413 414 415
	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
416

417
	dentry = d_lookup(parent, &filename);
B
Bryan Schumaker 已提交
418 419 420 421 422 423 424 425 426 427 428
	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);
429 430 431
	if (dentry == NULL)
		return;

B
Bryan Schumaker 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
	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
450 451 452
/* Perform conversion from xdr to cache array */
static
void nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *entry,
B
Bryan Schumaker 已提交
453
				void *xdr_page, struct page *page, unsigned int buflen)
L
Linus Torvalds 已提交
454
{
455 456
	struct xdr_stream stream;
	struct xdr_buf buf;
B
Bryan Schumaker 已提交
457
	__be32 *ptr = xdr_page;
B
Bryan Schumaker 已提交
458 459
	int status;
	struct nfs_cache_array *array;
460 461 462 463 464 465 466 467 468 469 470

	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
471 472 473 474 475 476

	do {
		status = xdr_decode(desc, entry, &stream);
		if (status != 0)
			break;

B
Bryan Schumaker 已提交
477
		if (nfs_readdir_add_to_array(entry, page) == -1)
L
Linus Torvalds 已提交
478
			break;
B
Bryan Schumaker 已提交
479 480
		if (desc->plus == 1)
			nfs_prime_dcache(desc->file->f_path.dentry, entry);
B
Bryan Schumaker 已提交
481 482 483 484 485 486 487
	} while (!entry->eof);

	if (status == -EBADCOOKIE && entry->eof) {
		array = nfs_readdir_get_array(page);
		array->eof_index = array->size - 1;
		status = 0;
		nfs_readdir_release_array(page);
L
Linus Torvalds 已提交
488
	}
B
Bryan Schumaker 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
}

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;
	}

524
	ptr = vm_map_ram(pages, npages, 0, PAGE_KERNEL);
B
Bryan Schumaker 已提交
525 526 527 528 529
	if (!IS_ERR_OR_NULL(ptr))
		return ptr;
out_freepages:
	nfs_readdir_free_pagearray(pages, i);
	return NULL;
L
Linus Torvalds 已提交
530 531
}

B
Bryan Schumaker 已提交
532 533
static
int nfs_readdir_xdr_to_array(nfs_readdir_descriptor_t *desc, struct page *page, struct inode *inode)
534
{
B
Bryan Schumaker 已提交
535 536
	struct page *pages[NFS_MAX_READDIR_PAGES];
	void *pages_ptr = NULL;
B
Bryan Schumaker 已提交
537 538 539 540
	struct nfs_entry entry;
	struct file	*file = desc->file;
	struct nfs_cache_array *array;
	int status = 0;
B
Bryan Schumaker 已提交
541
	unsigned int array_size = ARRAY_SIZE(pages);
B
Bryan Schumaker 已提交
542 543 544 545 546 547 548 549

	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;
550

B
Bryan Schumaker 已提交
551 552 553
	array = nfs_readdir_get_array(page);
	memset(array, 0, sizeof(struct nfs_cache_array));
	array->eof_index = -1;
554

B
Bryan Schumaker 已提交
555 556
	pages_ptr = nfs_readdir_large_page(pages, array_size);
	if (!pages_ptr)
B
Bryan Schumaker 已提交
557 558
		goto out_release_array;
	do {
B
Bryan Schumaker 已提交
559
		status = nfs_readdir_xdr_filler(pages, desc, &entry, file, inode);
560

B
Bryan Schumaker 已提交
561
		if (status < 0)
562
			break;
B
Bryan Schumaker 已提交
563
		nfs_readdir_page_filler(desc, &entry, pages_ptr, page, array_size * PAGE_SIZE);
B
Bryan Schumaker 已提交
564 565
	} while (array->eof_index < 0 && array->size < MAX_READDIR_ARRAY);

B
Bryan Schumaker 已提交
566
	nfs_readdir_free_large_page(pages_ptr, pages, array_size);
B
Bryan Schumaker 已提交
567 568 569 570 571
out_release_array:
	nfs_readdir_release_array(page);
out:
	nfs_free_fattr(entry.fattr);
	nfs_free_fhandle(entry.fh);
572 573 574 575
	return status;
}

/*
B
Bryan Schumaker 已提交
576 577 578 579
 * 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
580
 */
B
Bryan Schumaker 已提交
581 582
static
int nfs_readdir_filler(nfs_readdir_descriptor_t *desc, struct page* page)
L
Linus Torvalds 已提交
583
{
584
	struct inode	*inode = desc->file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
585

586
	if (nfs_readdir_xdr_to_array(desc, page, inode) < 0)
B
Bryan Schumaker 已提交
587 588
		goto error;
	SetPageUptodate(page);
L
Linus Torvalds 已提交
589

B
Bryan Schumaker 已提交
590 591 592
	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
593
	}
B
Bryan Schumaker 已提交
594 595 596 597 598 599
	unlock_page(page);
	return 0;
 error:
	unlock_page(page);
	return -EIO;
}
L
Linus Torvalds 已提交
600

B
Bryan Schumaker 已提交
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616
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)
{
	struct page *page;
	page = read_cache_page(desc->file->f_path.dentry->d_inode->i_mapping,
			desc->page_index, (filler_t *)nfs_readdir_filler, desc);
	if (IS_ERR(page))
		desc->eof = 1;
	return page;
L
Linus Torvalds 已提交
617 618 619
}

/*
B
Bryan Schumaker 已提交
620
 * Returns 0 if desc->dir_cookie was found on page desc->page_index
L
Linus Torvalds 已提交
621
 */
B
Bryan Schumaker 已提交
622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
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
639 640 641
static inline
int readdir_search_pagecache(nfs_readdir_descriptor_t *desc)
{
B
Bryan Schumaker 已提交
642 643 644 645
	int res = -EAGAIN;

	while (1) {
		res = find_cache_page(desc);
L
Linus Torvalds 已提交
646 647
		if (res != -EAGAIN)
			break;
B
Bryan Schumaker 已提交
648
		desc->page_index++;
L
Linus Torvalds 已提交
649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665
	}
	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
666 667 668 669 670 671 672 673 674 675
	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);

	for (i = desc->cache_entry_index; i < array->size; i++) {
		d_type = DT_UNKNOWN;
L
Linus Torvalds 已提交
676

B
Bryan Schumaker 已提交
677 678 679
		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
680 681
		if (res < 0)
			break;
682
		file->f_pos++;
B
Bryan Schumaker 已提交
683 684 685 686 687 688 689
		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;
		if (i == array->eof_index) {
			desc->eof = 1;
L
Linus Torvalds 已提交
690 691 692
			break;
		}
	}
B
Bryan Schumaker 已提交
693 694 695

	nfs_readdir_release_array(desc->page);
	cache_page_release(desc);
L
Linus Torvalds 已提交
696 697
	if (dentry != NULL)
		dput(dentry);
C
Chuck Lever 已提交
698 699
	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
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
	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
721
	struct inode *inode = desc->file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
722

C
Chuck Lever 已提交
723 724
	dfprintk(DIRCACHE, "NFS: uncached_readdir() searching for cookie %Lu\n",
			(unsigned long long)*desc->dir_cookie);
L
Linus Torvalds 已提交
725 726 727 728 729 730

	page = alloc_page(GFP_HIGHUSER);
	if (!page) {
		status = -ENOMEM;
		goto out;
	}
B
Bryan Schumaker 已提交
731 732

	if (nfs_readdir_xdr_to_array(desc, page, inode) == -1) {
L
Linus Torvalds 已提交
733 734
		status = -EIO;
		goto out_release;
B
Bryan Schumaker 已提交
735
	}
L
Linus Torvalds 已提交
736

737
	desc->page_index = 0;
B
Bryan Schumaker 已提交
738
	desc->page = page;
L
Linus Torvalds 已提交
739 740 741
	status = nfs_do_filldir(desc, dirent, filldir);

 out:
C
Chuck Lever 已提交
742
	dfprintk(DIRCACHE, "NFS: %s: returns %d\n",
743
			__func__, status);
L
Linus Torvalds 已提交
744 745
	return status;
 out_release:
B
Bryan Schumaker 已提交
746
	cache_page_release(desc);
L
Linus Torvalds 已提交
747 748 749
	goto out;
}

750 751 752
/* 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
753 754 755
 */
static int nfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
756
	struct dentry	*dentry = filp->f_path.dentry;
L
Linus Torvalds 已提交
757 758 759
	struct inode	*inode = dentry->d_inode;
	nfs_readdir_descriptor_t my_desc,
			*desc = &my_desc;
760
	int res = -ENOMEM;
L
Linus Torvalds 已提交
761

C
Chuck Lever 已提交
762
	dfprintk(FILE, "NFS: readdir(%s/%s) starting at cookie %llu\n",
C
Chuck Lever 已提交
763 764
			dentry->d_parent->d_name.name, dentry->d_name.name,
			(long long)filp->f_pos);
C
Chuck Lever 已提交
765 766
	nfs_inc_stats(inode, NFSIOS_VFSGETDENTS);

L
Linus Torvalds 已提交
767
	/*
768
	 * filp->f_pos points to the dirent entry number.
769
	 * *desc->dir_cookie has the cookie for the next entry. We have
770 771
	 * to either find the entry with the appropriate number or
	 * revalidate the cookie.
L
Linus Torvalds 已提交
772 773 774 775
	 */
	memset(desc, 0, sizeof(*desc));

	desc->file = filp;
776
	desc->dir_cookie = &nfs_file_open_context(filp)->dir_cookie;
L
Linus Torvalds 已提交
777 778 779
	desc->decode = NFS_PROTO(inode)->decode_dirent;
	desc->plus = NFS_USE_READDIRPLUS(inode);

T
Trond Myklebust 已提交
780
	nfs_block_sillyrename(dentry);
781
	res = nfs_revalidate_mapping(inode, filp->f_mapping);
T
Trond Myklebust 已提交
782 783 784
	if (res < 0)
		goto out;

B
Bryan Schumaker 已提交
785
	while (desc->eof != 1) {
L
Linus Torvalds 已提交
786
		res = readdir_search_pagecache(desc);
787

L
Linus Torvalds 已提交
788 789
		if (res == -EBADCOOKIE) {
			/* This means either end of directory */
B
Bryan Schumaker 已提交
790
			if (*desc->dir_cookie && desc->eof == 0) {
L
Linus Torvalds 已提交
791 792 793 794 795 796 797 798 799
				/* 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
800
			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
L
Linus Torvalds 已提交
801
			nfs_zap_caches(inode);
802
			desc->page_index = 0;
L
Linus Torvalds 已提交
803
			desc->plus = 0;
B
Bryan Schumaker 已提交
804
			desc->eof = 0;
L
Linus Torvalds 已提交
805 806 807 808 809 810 811 812 813 814 815
			continue;
		}
		if (res < 0)
			break;

		res = nfs_do_filldir(desc, dirent, filldir);
		if (res < 0) {
			res = 0;
			break;
		}
	}
T
Trond Myklebust 已提交
816
out:
T
Trond Myklebust 已提交
817
	nfs_unblock_sillyrename(dentry);
C
Chuck Lever 已提交
818 819
	if (res > 0)
		res = 0;
820
	dfprintk(FILE, "NFS: readdir(%s/%s) returns %d\n",
C
Chuck Lever 已提交
821 822 823
			dentry->d_parent->d_name.name, dentry->d_name.name,
			res);
	return res;
L
Linus Torvalds 已提交
824 825
}

826
static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int origin)
827
{
828 829 830
	struct dentry *dentry = filp->f_path.dentry;
	struct inode *inode = dentry->d_inode;

C
Chuck Lever 已提交
831
	dfprintk(FILE, "NFS: llseek dir(%s/%s, %lld, %d)\n",
832 833 834 835 836
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			offset, origin);

	mutex_lock(&inode->i_mutex);
837 838 839 840 841 842 843 844 845 846 847 848
	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;
849
		nfs_file_open_context(filp)->dir_cookie = 0;
850 851
	}
out:
852
	mutex_unlock(&inode->i_mutex);
853 854 855
	return offset;
}

L
Linus Torvalds 已提交
856 857 858 859
/*
 * All directory operations under NFS are synchronous, so fsync()
 * is a dummy operation.
 */
860
static int nfs_fsync_dir(struct file *filp, int datasync)
L
Linus Torvalds 已提交
861
{
862 863
	struct dentry *dentry = filp->f_path.dentry;

C
Chuck Lever 已提交
864
	dfprintk(FILE, "NFS: fsync dir(%s/%s) datasync %d\n",
C
Chuck Lever 已提交
865 866 867
			dentry->d_parent->d_name.name, dentry->d_name.name,
			datasync);

868
	nfs_inc_stats(dentry->d_inode, NFSIOS_VFSFSYNC);
L
Linus Torvalds 已提交
869 870 871
	return 0;
}

872 873 874 875 876 877 878 879 880 881 882 883
/**
 * 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)
{
884
	NFS_I(dir)->cache_change_attribute++;
885 886
}

L
Linus Torvalds 已提交
887 888 889 890 891
/*
 * 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.
 */
892
static int nfs_check_verifier(struct inode *dir, struct dentry *dentry)
L
Linus Torvalds 已提交
893 894 895
{
	if (IS_ROOT(dentry))
		return 1;
896 897
	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONE)
		return 0;
898 899 900 901 902 903 904 905
	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
906 907
}

908 909 910 911 912 913 914 915 916 917 918 919 920 921
/*
 * 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;
}

922 923 924 925 926 927 928 929
/*
 * 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;
930
	return nd && nfs_lookup_check_intent(nd, LOOKUP_EXCL);
931 932
}

933 934 935 936 937 938 939 940
/*
 * 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
941 942 943 944 945
static inline
int nfs_lookup_verify_inode(struct inode *inode, struct nameidata *nd)
{
	struct nfs_server *server = NFS_SERVER(inode);

946 947
	if (test_bit(NFS_INO_MOUNTPOINT, &NFS_I(inode)->flags))
		return 0;
L
Linus Torvalds 已提交
948 949
	if (nd != NULL) {
		/* VFS wants an on-the-wire revalidation */
950
		if (nd->flags & LOOKUP_REVAL)
L
Linus Torvalds 已提交
951 952
			goto out_force;
		/* This is an open(2) */
953
		if (nfs_lookup_check_intent(nd, LOOKUP_OPEN) != 0 &&
954 955 956
				!(server->flags & NFS_MOUNT_NOCTO) &&
				(S_ISREG(inode->i_mode) ||
				 S_ISDIR(inode->i_mode)))
L
Linus Torvalds 已提交
957
			goto out_force;
958
		return 0;
L
Linus Torvalds 已提交
959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976
	}
	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 */
977
	if (nd != NULL && nfs_lookup_check_intent(nd, LOOKUP_CREATE) != 0)
L
Linus Torvalds 已提交
978
		return 0;
979 980
	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONEG)
		return 1;
L
Linus Torvalds 已提交
981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999
	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;
1000 1001
	struct nfs_fh *fhandle = NULL;
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
1002 1003 1004 1005
	int error;

	parent = dget_parent(dentry);
	dir = parent->d_inode;
C
Chuck Lever 已提交
1006
	nfs_inc_stats(dir, NFSIOS_DENTRYREVALIDATE);
L
Linus Torvalds 已提交
1007 1008 1009 1010 1011 1012 1013 1014 1015
	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
1016
		dfprintk(LOOKUPCACHE, "%s: %s/%s has dud inode\n",
1017
				__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1018
				dentry->d_name.name);
L
Linus Torvalds 已提交
1019 1020 1021
		goto out_bad;
	}

1022 1023 1024
	if (nfs_have_delegation(inode, FMODE_READ))
		goto out_set_verifier;

L
Linus Torvalds 已提交
1025
	/* Force a full look up iff the parent directory has changed */
1026
	if (!nfs_is_exclusive_create(dir, nd) && nfs_check_verifier(dir, dentry)) {
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031 1032 1033 1034
		if (nfs_lookup_verify_inode(inode, nd))
			goto out_zap_parent;
		goto out_valid;
	}

	if (NFS_STALE(inode))
		goto out_bad;

1035 1036 1037 1038 1039 1040 1041
	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
1042 1043
	if (error)
		goto out_bad;
1044
	if (nfs_compare_fh(NFS_FH(inode), fhandle))
L
Linus Torvalds 已提交
1045
		goto out_bad;
1046
	if ((error = nfs_refresh_inode(inode, fattr)) != 0)
L
Linus Torvalds 已提交
1047 1048
		goto out_bad;

1049 1050
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
1051
out_set_verifier:
1052
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1053 1054
 out_valid:
	dput(parent);
C
Chuck Lever 已提交
1055
	dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) is valid\n",
1056
			__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1057
			dentry->d_name.name);
L
Linus Torvalds 已提交
1058 1059 1060 1061
	return 1;
out_zap_parent:
	nfs_zap_caches(dir);
 out_bad:
1062
	nfs_mark_for_revalidate(dir);
L
Linus Torvalds 已提交
1063 1064 1065 1066 1067 1068
	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;
1069 1070
		if (dentry->d_flags & DCACHE_DISCONNECTED)
			goto out_valid;
L
Linus Torvalds 已提交
1071 1072 1073
		shrink_dcache_parent(dentry);
	}
	d_drop(dentry);
1074 1075
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
L
Linus Torvalds 已提交
1076
	dput(parent);
C
Chuck Lever 已提交
1077
	dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) is invalid\n",
1078
			__func__, dentry->d_parent->d_name.name,
C
Chuck Lever 已提交
1079
			dentry->d_name.name);
L
Linus Torvalds 已提交
1080
	return 0;
1081 1082 1083 1084 1085 1086 1087 1088
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
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
}

/*
 * 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);

1100 1101 1102 1103
	/* 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
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
	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;

}

1117 1118 1119 1120 1121 1122 1123 1124
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
1125 1126 1127 1128 1129 1130
/*
 * 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)
{
1131 1132 1133 1134
	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
1135
	if (dentry->d_flags & DCACHE_NFSFS_RENAMED) {
1136
		drop_nlink(inode);
1137
		nfs_complete_unlink(dentry, inode);
L
Linus Torvalds 已提交
1138 1139 1140 1141
	}
	iput(inode);
}

A
Al Viro 已提交
1142
const struct dentry_operations nfs_dentry_operations = {
L
Linus Torvalds 已提交
1143 1144 1145 1146 1147 1148 1149 1150
	.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
1151
	struct dentry *parent;
L
Linus Torvalds 已提交
1152
	struct inode *inode = NULL;
1153 1154
	struct nfs_fh *fhandle = NULL;
	struct nfs_fattr *fattr = NULL;
L
Linus Torvalds 已提交
1155 1156 1157 1158
	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
1159
	nfs_inc_stats(dir, NFSIOS_VFSLOOKUP);
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164 1165 1166

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

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

1167 1168 1169 1170 1171 1172 1173
	/*
	 * 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;
1174
		goto out;
1175
	}
L
Linus Torvalds 已提交
1176

1177 1178 1179 1180 1181 1182
	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
1183 1184 1185
	parent = dentry->d_parent;
	/* Protect against concurrent sillydeletes */
	nfs_block_sillyrename(parent);
1186
	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
L
Linus Torvalds 已提交
1187 1188 1189 1190
	if (error == -ENOENT)
		goto no_entry;
	if (error < 0) {
		res = ERR_PTR(error);
T
Trond Myklebust 已提交
1191
		goto out_unblock_sillyrename;
L
Linus Torvalds 已提交
1192
	}
1193
	inode = nfs_fhget(dentry->d_sb, fhandle, fattr);
1194 1195
	res = (struct dentry *)inode;
	if (IS_ERR(res))
T
Trond Myklebust 已提交
1196
		goto out_unblock_sillyrename;
1197

L
Linus Torvalds 已提交
1198
no_entry:
1199
	res = d_materialise_unique(dentry, inode);
1200 1201
	if (res != NULL) {
		if (IS_ERR(res))
T
Trond Myklebust 已提交
1202
			goto out_unblock_sillyrename;
L
Linus Torvalds 已提交
1203
		dentry = res;
1204
	}
L
Linus Torvalds 已提交
1205
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
T
Trond Myklebust 已提交
1206 1207
out_unblock_sillyrename:
	nfs_unblock_sillyrename(parent);
L
Linus Torvalds 已提交
1208
out:
1209 1210
	nfs_free_fattr(fattr);
	nfs_free_fhandle(fhandle);
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215 1216
	return res;
}

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

A
Al Viro 已提交
1217
const struct dentry_operations nfs4_dentry_operations = {
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222
	.d_revalidate	= nfs_open_revalidate,
	.d_delete	= nfs_dentry_delete,
	.d_iput		= nfs_dentry_iput,
};

1223 1224 1225 1226
/*
 * 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
1227
static int is_atomic_open(struct nameidata *nd)
L
Linus Torvalds 已提交
1228
{
1229
	if (nd == NULL || nfs_lookup_check_intent(nd, LOOKUP_OPEN) == 0)
L
Linus Torvalds 已提交
1230 1231 1232 1233 1234
		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? */
1235 1236
	if (__mnt_is_readonly(nd->path.mnt) &&
	    (nd->intent.open.flags & (O_CREAT|O_TRUNC|FMODE_WRITE)))
L
Linus Torvalds 已提交
1237 1238 1239 1240
		return 0;
	return 1;
}

1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
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
1290 1291
static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
{
1292 1293
	struct nfs_open_context *ctx;
	struct iattr attr;
L
Linus Torvalds 已提交
1294
	struct dentry *res = NULL;
1295
	struct inode *inode;
1296
	int open_flags;
1297
	int err;
L
Linus Torvalds 已提交
1298

C
Chuck Lever 已提交
1299 1300 1301
	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
1302
	/* Check that we are indeed trying to open this file */
T
Trond Myklebust 已提交
1303
	if (!is_atomic_open(nd))
L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309 1310 1311
		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;

1312 1313
	/* Let vfs_create() deal with O_EXCL. Instantiate, but don't hash
	 * the dentry. */
1314
	if (nd->flags & LOOKUP_EXCL) {
1315
		d_instantiate(dentry, NULL);
1316 1317
		goto out;
	}
L
Linus Torvalds 已提交
1318

1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
	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 {
1331
		open_flags &= ~(O_EXCL | O_CREAT);
1332 1333 1334
		attr.ia_valid = 0;
	}

L
Linus Torvalds 已提交
1335
	/* Open the file on the server */
1336
	nfs_block_sillyrename(dentry->d_parent);
1337
	inode = NFS_PROTO(dir)->open_context(dir, ctx, open_flags, &attr);
1338 1339
	if (IS_ERR(inode)) {
		nfs_unblock_sillyrename(dentry->d_parent);
1340
		put_nfs_open_context(ctx);
1341
		switch (PTR_ERR(inode)) {
L
Linus Torvalds 已提交
1342 1343
			/* Make a negative dentry */
			case -ENOENT:
1344
				d_add(dentry, NULL);
1345 1346
				res = NULL;
				goto out;
L
Linus Torvalds 已提交
1347
			/* This turned out not to be a regular file */
1348 1349
			case -ENOTDIR:
				goto no_open;
L
Linus Torvalds 已提交
1350 1351 1352
			case -ELOOP:
				if (!(nd->intent.open.flags & O_NOFOLLOW))
					goto no_open;
1353
			/* case -EISDIR: */
L
Linus Torvalds 已提交
1354 1355
			/* case -EINVAL: */
			default:
1356
				res = ERR_CAST(inode);
L
Linus Torvalds 已提交
1357 1358
				goto out;
		}
1359
	}
1360
	res = d_add_unique(dentry, inode);
1361
	nfs_unblock_sillyrename(dentry->d_parent);
1362 1363 1364
	if (res != NULL) {
		dput(ctx->path.dentry);
		ctx->path.dentry = dget(res);
L
Linus Torvalds 已提交
1365
		dentry = res;
1366
	}
1367 1368 1369 1370 1371 1372
	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
1373
out:
1374
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1375 1376 1377 1378 1379 1380 1381 1382 1383 1384
	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;
1385
	struct nfs_open_context *ctx;
L
Linus Torvalds 已提交
1386 1387
	int openflags, ret = 0;

1388
	if (!is_atomic_open(nd) || d_mountpoint(dentry))
T
Trond Myklebust 已提交
1389
		goto no_open;
1390

L
Linus Torvalds 已提交
1391 1392
	parent = dget_parent(dentry);
	dir = parent->d_inode;
1393

L
Linus Torvalds 已提交
1394 1395 1396
	/* We can't create new files in nfs_open_revalidate(), so we
	 * optimize away revalidation of negative dentries.
	 */
1397 1398 1399
	if (inode == NULL) {
		if (!nfs_neg_need_reval(dir, dentry, nd))
			ret = 1;
L
Linus Torvalds 已提交
1400
		goto out;
1401 1402
	}

L
Linus Torvalds 已提交
1403 1404
	/* NFS only supports OPEN on regular files */
	if (!S_ISREG(inode->i_mode))
T
Trond Myklebust 已提交
1405
		goto no_open_dput;
L
Linus Torvalds 已提交
1406 1407 1408
	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
1409
		goto no_open_dput;
L
Linus Torvalds 已提交
1410
	/* We can't create new files, or truncate existing ones here */
1411
	openflags &= ~(O_CREAT|O_EXCL|O_TRUNC);
L
Linus Torvalds 已提交
1412

1413 1414 1415 1416
	ctx = nameidata_to_nfs_open_context(dentry, nd);
	ret = PTR_ERR(ctx);
	if (IS_ERR(ctx))
		goto out;
L
Linus Torvalds 已提交
1417
	/*
1418
	 * Note: we're not holding inode->i_mutex and so may be racing with
L
Linus Torvalds 已提交
1419 1420 1421
	 * operations that change the directory. We therefore save the
	 * change attribute *before* we do the RPC call.
	 */
1422
	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, NULL);
1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
	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);
1437
	if (inode != dentry->d_inode)
1438
		goto out_drop;
1439 1440 1441 1442 1443

	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
1444 1445 1446
out:
	dput(parent);
	return ret;
1447 1448 1449 1450 1451 1452 1453
out_drop:
	d_drop(dentry);
	ret = 0;
out_put_ctx:
	put_nfs_open_context(ctx);
	goto out;

T
Trond Myklebust 已提交
1454
no_open_dput:
L
Linus Torvalds 已提交
1455
	dput(parent);
T
Trond Myklebust 已提交
1456
no_open:
L
Linus Torvalds 已提交
1457 1458
	return nfs_lookup_revalidate(dentry, nd);
}
1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479

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))
1480
			goto out_err_drop;
1481 1482 1483 1484 1485
	}

	error = NFS_PROTO(dir)->create(dir, dentry, &attr, open_flags, ctx);
	if (error != 0)
		goto out_put_ctx;
1486 1487 1488 1489 1490
	if (ctx != NULL) {
		error = nfs_intent_set_file(nd, ctx);
		if (error < 0)
			goto out_err;
	}
1491 1492 1493 1494
	return 0;
out_put_ctx:
	if (ctx != NULL)
		put_nfs_open_context(ctx);
1495
out_err_drop:
1496
	d_drop(dentry);
1497
out_err:
1498 1499 1500
	return error;
}

L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508
#endif /* CONFIG_NFSV4 */

/*
 * Code common to create, mkdir, and mknod.
 */
int nfs_instantiate(struct dentry *dentry, struct nfs_fh *fhandle,
				struct nfs_fattr *fattr)
{
1509 1510
	struct dentry *parent = dget_parent(dentry);
	struct inode *dir = parent->d_inode;
L
Linus Torvalds 已提交
1511 1512 1513
	struct inode *inode;
	int error = -EACCES;

1514 1515
	d_drop(dentry);

L
Linus Torvalds 已提交
1516 1517
	/* We may have been initialized further down */
	if (dentry->d_inode)
1518
		goto out;
L
Linus Torvalds 已提交
1519 1520 1521
	if (fhandle->size == 0) {
		error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
		if (error)
1522
			goto out_error;
L
Linus Torvalds 已提交
1523
	}
1524
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
1525 1526
	if (!(fattr->valid & NFS_ATTR_FATTR)) {
		struct nfs_server *server = NFS_SB(dentry->d_sb);
1527
		error = server->nfs_client->rpc_ops->getattr(server, fhandle, fattr);
L
Linus Torvalds 已提交
1528
		if (error < 0)
1529
			goto out_error;
L
Linus Torvalds 已提交
1530 1531
	}
	inode = nfs_fhget(dentry->d_sb, fhandle, fattr);
1532 1533
	error = PTR_ERR(inode);
	if (IS_ERR(inode))
1534 1535 1536 1537
		goto out_error;
	d_add(dentry, inode);
out:
	dput(parent);
L
Linus Torvalds 已提交
1538
	return 0;
1539 1540 1541 1542
out_error:
	nfs_mark_for_revalidate(dir);
	dput(parent);
	return error;
L
Linus Torvalds 已提交
1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
}

/*
 * 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
1557 1558
	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
1559 1560 1561 1562

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

1563
	error = NFS_PROTO(dir)->create(dir, dentry, &attr, 0, NULL);
L
Linus Torvalds 已提交
1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
	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
1581 1582
	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
1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606

	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
1607 1608
	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
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621

	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;
}

1622 1623 1624 1625 1626 1627
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
1628 1629 1630 1631
static int nfs_rmdir(struct inode *dir, struct dentry *dentry)
{
	int error;

C
Chuck Lever 已提交
1632 1633
	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
1634 1635 1636 1637

	error = NFS_PROTO(dir)->rmdir(dir, &dentry->d_name);
	/* Ensure the VFS deletes this inode */
	if (error == 0 && dentry->d_inode != NULL)
1638
		clear_nlink(dentry->d_inode);
1639 1640
	else if (error == -ENOENT)
		nfs_dentry_handle_enoent(dentry);
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667

	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) {
1668
		nfs_inode_return_delegation(inode);
L
Linus Torvalds 已提交
1669 1670 1671
		error = NFS_PROTO(dir)->remove(dir, &dentry->d_name);
		/* The VFS may want to delete this inode */
		if (error == 0)
1672
			nfs_drop_nlink(inode);
1673
		nfs_mark_for_revalidate(inode);
L
Linus Torvalds 已提交
1674 1675
	} else
		error = NFS_PROTO(dir)->remove(dir, &dentry->d_name);
1676 1677
	if (error == -ENOENT)
		nfs_dentry_handle_enoent(dentry);
L
Linus Torvalds 已提交
1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
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
1700 1701
		/* Start asynchronous writeout of the inode */
		write_inode_now(dentry->d_inode, 0);
L
Linus Torvalds 已提交
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711
		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);
1712
	if (!error || error == -ENOENT) {
L
Linus Torvalds 已提交
1713 1714 1715 1716 1717 1718
		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
	} else if (need_rehash)
		d_rehash(dentry);
	return error;
}

1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
/*
 * 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
1735
{
1736 1737 1738
	struct pagevec lru_pvec;
	struct page *page;
	char *kaddr;
L
Linus Torvalds 已提交
1739
	struct iattr attr;
1740
	unsigned int pathlen = strlen(symname);
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745
	int error;

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

1746 1747
	if (pathlen > PAGE_SIZE)
		return -ENAMETOOLONG;
L
Linus Torvalds 已提交
1748

1749 1750
	attr.ia_mode = S_IFLNK | S_IRWXUGO;
	attr.ia_valid = ATTR_MODE;
L
Linus Torvalds 已提交
1751

1752
	page = alloc_page(GFP_HIGHUSER);
1753
	if (!page)
1754 1755 1756 1757 1758 1759 1760 1761
		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);

1762
	error = NFS_PROTO(dir)->symlink(dir, dentry, page, pathlen, &attr);
1763 1764 1765 1766
	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
1767
		d_drop(dentry);
1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778
		__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)) {
1779
		pagevec_add(&lru_pvec, page);
1780
		pagevec_lru_add_file(&lru_pvec);
1781 1782 1783 1784 1785 1786
		SetPageUptodate(page);
		unlock_page(page);
	} else
		__free_page(page);

	return 0;
L
Linus Torvalds 已提交
1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
}

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);

1799 1800
	nfs_inode_return_delegation(inode);

1801
	d_drop(dentry);
L
Linus Torvalds 已提交
1802
	error = NFS_PROTO(dir)->link(inode, dir, &dentry->d_name);
1803
	if (error == 0) {
A
Al Viro 已提交
1804
		ihold(inode);
1805
		d_add(dentry, inode);
1806
	}
L
Linus Torvalds 已提交
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
	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
1848 1849 1850 1851
	 * 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
1852
	 */
1853 1854 1855 1856 1857 1858 1859 1860 1861
	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
1862

1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
		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);
1874
			if (err)
1875
				goto out;
1876 1877

			new_dentry = dentry;
O
OGAWA Hirofumi 已提交
1878
			rehash = NULL;
1879
			new_inode = NULL;
1880
		}
1881
	}
L
Linus Torvalds 已提交
1882

1883
	nfs_inode_return_delegation(old_inode);
1884
	if (new_inode != NULL)
1885
		nfs_inode_return_delegation(new_inode);
L
Linus Torvalds 已提交
1886 1887 1888

	error = NFS_PROTO(old_dir)->rename(old_dir, &old_dentry->d_name,
					   new_dir, &new_dentry->d_name);
1889
	nfs_mark_for_revalidate(old_inode);
L
Linus Torvalds 已提交
1890 1891 1892 1893
out:
	if (rehash)
		d_rehash(rehash);
	if (!error) {
1894 1895
		if (new_inode != NULL)
			nfs_drop_nlink(new_inode);
1896
		d_move(old_dentry, new_dentry);
1897 1898
		nfs_set_verifier(new_dentry,
					nfs_save_change_attribute(new_dir));
1899 1900
	} else if (error == -ENOENT)
		nfs_dentry_handle_enoent(old_dentry);
L
Linus Torvalds 已提交
1901 1902 1903 1904 1905 1906 1907

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

1908 1909 1910 1911
static DEFINE_SPINLOCK(nfs_access_lru_lock);
static LIST_HEAD(nfs_access_lru_list);
static atomic_long_t nfs_access_nr_entries;

1912 1913 1914 1915
static void nfs_access_free_entry(struct nfs_access_entry *entry)
{
	put_rpccred(entry->cred);
	kfree(entry);
1916 1917 1918
	smp_mb__before_atomic_dec();
	atomic_long_dec(&nfs_access_nr_entries);
	smp_mb__after_atomic_dec();
1919 1920
}

1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931
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);
	}
}

1932
int nfs_access_cache_shrinker(struct shrinker *shrink, int nr_to_scan, gfp_t gfp_mask)
1933 1934
{
	LIST_HEAD(head);
1935
	struct nfs_inode *nfsi, *next;
1936 1937
	struct nfs_access_entry *cache;

1938 1939
	if ((gfp_mask & GFP_KERNEL) != GFP_KERNEL)
		return (nr_to_scan == 0) ? 0 : -1;
1940

1941
	spin_lock(&nfs_access_lru_lock);
1942
	list_for_each_entry_safe(nfsi, next, &nfs_access_lru_list, access_cache_inode_lru) {
1943 1944 1945 1946
		struct inode *inode;

		if (nr_to_scan-- == 0)
			break;
1947
		inode = &nfsi->vfs_inode;
1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
		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);
1961
			smp_mb__before_clear_bit();
1962
			clear_bit(NFS_INO_ACL_LRU_SET, &nfsi->flags);
1963
			smp_mb__after_clear_bit();
1964
		}
1965
		spin_unlock(&inode->i_lock);
1966 1967
	}
	spin_unlock(&nfs_access_lru_lock);
1968
	nfs_access_free_list(&head);
1969 1970 1971
	return (atomic_long_read(&nfs_access_nr_entries) / 100) * sysctl_vfs_cache_pressure;
}

1972
static void __nfs_access_zap_cache(struct nfs_inode *nfsi, struct list_head *head)
L
Linus Torvalds 已提交
1973
{
1974
	struct rb_root *root_node = &nfsi->access_cache;
1975
	struct rb_node *n;
1976 1977 1978 1979 1980 1981
	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);
1982
		list_move(&entry->lru, head);
1983 1984
	}
	nfsi->cache_validity &= ~NFS_INO_INVALID_ACCESS;
L
Linus Torvalds 已提交
1985 1986
}

1987
void nfs_access_zap_cache(struct inode *inode)
L
Linus Torvalds 已提交
1988
{
1989 1990 1991 1992
	LIST_HEAD(head);

	if (test_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags) == 0)
		return;
1993
	/* Remove from global LRU init */
1994 1995
	spin_lock(&nfs_access_lru_lock);
	if (test_and_clear_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags))
1996 1997
		list_del_init(&NFS_I(inode)->access_cache_inode_lru);

1998
	spin_lock(&inode->i_lock);
1999 2000 2001 2002
	__nfs_access_zap_cache(NFS_I(inode), &head);
	spin_unlock(&inode->i_lock);
	spin_unlock(&nfs_access_lru_lock);
	nfs_access_free_list(&head);
2003
}
L
Linus Torvalds 已提交
2004

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
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
2019
	}
2020 2021 2022
	return NULL;
}

2023
static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res)
2024 2025 2026 2027 2028
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_access_entry *cache;
	int err = -ENOENT;

2029
	spin_lock(&inode->i_lock);
2030 2031 2032 2033 2034
	if (nfsi->cache_validity & NFS_INO_INVALID_ACCESS)
		goto out_zap;
	cache = nfs_access_search_rbtree(inode, cred);
	if (cache == NULL)
		goto out;
2035
	if (!nfs_have_delegated_attributes(inode) &&
2036
	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
2037 2038 2039 2040
		goto out_stale;
	res->jiffies = cache->jiffies;
	res->cred = cache->cred;
	res->mask = cache->mask;
2041
	list_move_tail(&cache->lru, &nfsi->access_cache_entry_lru);
2042 2043 2044 2045 2046 2047
	err = 0;
out:
	spin_unlock(&inode->i_lock);
	return err;
out_stale:
	rb_erase(&cache->rb_node, &nfsi->access_cache);
2048
	list_del(&cache->lru);
2049 2050 2051 2052
	spin_unlock(&inode->i_lock);
	nfs_access_free_entry(cache);
	return -ENOENT;
out_zap:
2053 2054
	spin_unlock(&inode->i_lock);
	nfs_access_zap_cache(inode);
2055 2056 2057 2058 2059
	return -ENOENT;
}

static void nfs_access_add_rbtree(struct inode *inode, struct nfs_access_entry *set)
{
2060 2061
	struct nfs_inode *nfsi = NFS_I(inode);
	struct rb_root *root_node = &nfsi->access_cache;
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079
	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);
2080
	list_add_tail(&set->lru, &nfsi->access_cache_entry_lru);
2081
	spin_unlock(&inode->i_lock);
2082 2083 2084
	return;
found:
	rb_replace_node(parent, &set->rb_node, root_node);
2085 2086
	list_add_tail(&set->lru, &nfsi->access_cache_entry_lru);
	list_del(&entry->lru);
2087 2088 2089 2090
	spin_unlock(&inode->i_lock);
	nfs_access_free_entry(entry);
}

2091
static void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set)
2092 2093 2094 2095 2096
{
	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
2097
	cache->jiffies = set->jiffies;
2098
	cache->cred = get_rpccred(set->cred);
L
Linus Torvalds 已提交
2099
	cache->mask = set->mask;
2100 2101

	nfs_access_add_rbtree(inode, cache);
2102 2103 2104 2105 2106 2107 2108

	/* 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 */
2109
	if (!test_bit(NFS_INO_ACL_LRU_SET, &NFS_I(inode)->flags)) {
2110
		spin_lock(&nfs_access_lru_lock);
2111 2112 2113
		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);
2114 2115
		spin_unlock(&nfs_access_lru_lock);
	}
L
Linus Torvalds 已提交
2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
}

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);
2132 2133 2134 2135 2136 2137
	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
2138
		return status;
2139
	}
L
Linus Torvalds 已提交
2140 2141
	nfs_access_add_cache(inode, &cache);
out:
2142
	if ((mask & ~cache.mask & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
L
Linus Torvalds 已提交
2143 2144 2145 2146
		return 0;
	return -EACCES;
}

2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164
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));
}

2165
int nfs_permission(struct inode *inode, int mask)
L
Linus Torvalds 已提交
2166 2167 2168 2169
{
	struct rpc_cred *cred;
	int res = 0;

C
Chuck Lever 已提交
2170 2171
	nfs_inc_stats(inode, NFSIOS_VFSACCESS);

2172
	if ((mask & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
L
Linus Torvalds 已提交
2173 2174
		goto out;
	/* Is this sys_access() ? */
E
Eric Paris 已提交
2175
	if (mask & (MAY_ACCESS | MAY_CHDIR))
L
Linus Torvalds 已提交
2176 2177 2178 2179 2180 2181 2182 2183
		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)
2184 2185
					&& (mask & MAY_OPEN)
					&& !(mask & MAY_EXEC))
L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200
				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;

2201
	cred = rpc_lookup_cred();
L
Linus Torvalds 已提交
2202 2203 2204 2205 2206 2207
	if (!IS_ERR(cred)) {
		res = nfs_do_access(inode, cred, mask);
		put_rpccred(cred);
	} else
		res = PTR_ERR(cred);
out:
2208 2209 2210
	if (!res && (mask & MAY_EXEC) && !execute_ok(inode))
		res = -EACCES;

C
Chuck Lever 已提交
2211 2212
	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
2213 2214 2215 2216 2217
	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
2218
	goto out;
L
Linus Torvalds 已提交
2219 2220 2221 2222 2223 2224 2225 2226
}

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