vfs.c 53.8 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
/*
 * File operations used by nfsd. Some of these have been ripped from
 * other parts of the kernel because they weren't exported, others
 * are partial duplicates with added or changed functionality.
 *
 * Note that several functions dget() the dentry upon which they want
 * to act, most notably those that create directory entries. Response
 * dentry's are dput()'d if necessary in the release callback.
 * So if you notice code paths that apparently fail to dput() the
 * dentry, don't worry--they have been taken care of.
 *
 * Copyright (C) 1995-1999 Olaf Kirch <okir@monad.swb.de>
 * Zerocpy NFS support (C) 2002 Hirokazu Takahashi <taka@valinux.co.jp>
 */

#include <linux/fs.h>
#include <linux/file.h>
18
#include <linux/splice.h>
L
Linus Torvalds 已提交
19 20 21
#include <linux/fcntl.h>
#include <linux/namei.h>
#include <linux/delay.h>
R
Robert Love 已提交
22
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
23 24
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
25 26
#include <linux/jhash.h>
#include <linux/ima.h>
27
#include <linux/slab.h>
28
#include <asm/uaccess.h>
29 30
#include <linux/exportfs.h>
#include <linux/writeback.h>
31 32 33 34 35

#ifdef CONFIG_NFSD_V3
#include "xdr3.h"
#endif /* CONFIG_NFSD_V3 */

36
#ifdef CONFIG_NFSD_V4
37 38
#include "acl.h"
#include "idmap.h"
L
Linus Torvalds 已提交
39 40
#endif /* CONFIG_NFSD_V4 */

41 42
#include "nfsd.h"
#include "vfs.h"
L
Linus Torvalds 已提交
43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

#define NFSDDBG_FACILITY		NFSDDBG_FILEOP


/*
 * This is a cache of readahead params that help us choose the proper
 * readahead strategy. Initially, we set all readahead parameters to 0
 * and let the VFS handle things.
 * If you increase the number of cached files very much, you'll need to
 * add a hash table here.
 */
struct raparms {
	struct raparms		*p_next;
	unsigned int		p_count;
	ino_t			p_ino;
	dev_t			p_dev;
	int			p_set;
	struct file_ra_state	p_ra;
61
	unsigned int		p_hindex;
L
Linus Torvalds 已提交
62 63
};

64 65 66 67 68 69 70 71 72
struct raparm_hbucket {
	struct raparms		*pb_head;
	spinlock_t		pb_lock;
} ____cacheline_aligned_in_smp;

#define RAPARM_HASH_BITS	4
#define RAPARM_HASH_SIZE	(1<<RAPARM_HASH_BITS)
#define RAPARM_HASH_MASK	(RAPARM_HASH_SIZE-1)
static struct raparm_hbucket	raparm_hash[RAPARM_HASH_SIZE];
L
Linus Torvalds 已提交
73 74 75 76

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
77
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
L
Linus Torvalds 已提交
78 79 80 81 82 83 84 85
 *  or nfs_ok having possibly changed *dpp and *expp
 */
int
nfsd_cross_mnt(struct svc_rqst *rqstp, struct dentry **dpp, 
		        struct svc_export **expp)
{
	struct svc_export *exp = *expp, *exp2 = NULL;
	struct dentry *dentry = *dpp;
A
Al Viro 已提交
86 87
	struct path path = {.mnt = mntget(exp->ex_path.mnt),
			    .dentry = dget(dentry)};
88
	int err = 0;
L
Linus Torvalds 已提交
89

90
	err = follow_down(&path);
91 92
	if (err < 0)
		goto out;
L
Linus Torvalds 已提交
93

A
Al Viro 已提交
94
	exp2 = rqst_exp_get_by_name(rqstp, &path);
L
Linus Torvalds 已提交
95
	if (IS_ERR(exp2)) {
96 97 98 99 100 101 102 103 104 105
		err = PTR_ERR(exp2);
		/*
		 * We normally allow NFS clients to continue
		 * "underneath" a mountpoint that is not exported.
		 * The exception is V4ROOT, where no traversal is ever
		 * allowed without an explicit export of the new
		 * directory.
		 */
		if (err == -ENOENT && !(exp->ex_flags & NFSEXP_V4ROOT))
			err = 0;
A
Al Viro 已提交
106
		path_put(&path);
L
Linus Torvalds 已提交
107 108
		goto out;
	}
109 110
	if (nfsd_v4client(rqstp) ||
		(exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
L
Linus Torvalds 已提交
111
		/* successfully crossed mount point */
A
Al Viro 已提交
112
		/*
A
Al Viro 已提交
113 114 115 116
		 * This is subtle: path.dentry is *not* on path.mnt
		 * at this point.  The only reason we are safe is that
		 * original mnt is pinned down by exp, so we should
		 * put path *before* putting exp
A
Al Viro 已提交
117
		 */
A
Al Viro 已提交
118 119
		*dpp = path.dentry;
		path.dentry = dentry;
A
Al Viro 已提交
120
		*expp = exp2;
A
Al Viro 已提交
121
		exp2 = exp;
L
Linus Torvalds 已提交
122
	}
A
Al Viro 已提交
123 124
	path_put(&path);
	exp_put(exp2);
L
Linus Torvalds 已提交
125 126 127 128
out:
	return err;
}

129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
static void follow_to_parent(struct path *path)
{
	struct dentry *dp;

	while (path->dentry == path->mnt->mnt_root && follow_up(path))
		;
	dp = dget_parent(path->dentry);
	dput(path->dentry);
	path->dentry = dp;
}

static int nfsd_lookup_parent(struct svc_rqst *rqstp, struct dentry *dparent, struct svc_export **exp, struct dentry **dentryp)
{
	struct svc_export *exp2;
	struct path path = {.mnt = mntget((*exp)->ex_path.mnt),
			    .dentry = dget(dparent)};

	follow_to_parent(&path);

	exp2 = rqst_exp_parent(rqstp, &path);
	if (PTR_ERR(exp2) == -ENOENT) {
		*dentryp = dget(dparent);
	} else if (IS_ERR(exp2)) {
		path_put(&path);
		return PTR_ERR(exp2);
	} else {
		*dentryp = dget(path.dentry);
		exp_put(*exp);
		*exp = exp2;
	}
	path_put(&path);
	return 0;
}

163 164 165 166
/*
 * For nfsd purposes, we treat V4ROOT exports as though there was an
 * export at *every* directory.
 */
167
int nfsd_mountpoint(struct dentry *dentry, struct svc_export *exp)
168 169 170
{
	if (d_mountpoint(dentry))
		return 1;
171 172
	if (nfsd4_is_junction(dentry))
		return 1;
173 174 175 176 177
	if (!(exp->ex_flags & NFSEXP_V4ROOT))
		return 0;
	return dentry->d_inode != NULL;
}

178
__be32
179
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
180
		   const char *name, unsigned int len,
181
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
182 183 184 185
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
186
	int			host_err;
L
Linus Torvalds 已提交
187 188 189 190 191 192 193 194 195 196 197

	dprintk("nfsd: nfsd_lookup(fh %s, %.*s)\n", SVCFH_fmt(fhp), len,name);

	dparent = fhp->fh_dentry;
	exp  = fhp->fh_export;
	exp_get(exp);

	/* Lookup the name, but don't follow links */
	if (isdotent(name, len)) {
		if (len==1)
			dentry = dget(dparent);
198
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
199
			dentry = dget_parent(dparent);
200
		else if (!EX_NOHIDE(exp) && !nfsd_v4client(rqstp))
L
Linus Torvalds 已提交
201 202 203
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
204 205
			host_err = nfsd_lookup_parent(rqstp, dparent, &exp, &dentry);
			if (host_err)
L
Linus Torvalds 已提交
206 207 208 209 210
				goto out_nfserr;
		}
	} else {
		fh_lock(fhp);
		dentry = lookup_one_len(name, dparent, len);
211
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
212 213 214 215 216
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
217
		if (nfsd_mountpoint(dentry, exp)) {
218
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
219 220 221 222 223
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
	*dentry_ret = dentry;
	*exp_ret = exp;
	return 0;

out_nfserr:
	exp_put(exp);
	return nfserrno(host_err);
}

/*
 * Look up one component of a pathname.
 * N.B. After this call _both_ fhp and resfh need an fh_put
 *
 * If the lookup would cross a mountpoint, and the mounted filesystem
 * is exported to the client with NFSEXP_NOHIDE, then the lookup is
 * accepted as it stands and the mounted directory is
 * returned. Otherwise the covered directory is returned.
 * NOTE: this mountpoint crossing is not supported properly by all
 *   clients and is explicitly disallowed for NFSv3
 *      NeilBrown <neilb@cse.unsw.edu.au>
 */
__be32
nfsd_lookup(struct svc_rqst *rqstp, struct svc_fh *fhp, const char *name,
247
				unsigned int len, struct svc_fh *resfh)
248 249 250 251 252
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

253 254 255
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
	if (err)
		return err;
256 257 258
	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
259 260 261
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
262 263 264 265 266 267 268
	/*
	 * Note: we compose the file handle now, but as the
	 * dentry may be negative, it may need to be updated.
	 */
	err = fh_compose(resfh, exp, dentry, fhp);
	if (!err && !dentry->d_inode)
		err = nfserr_noent;
269
out:
L
Linus Torvalds 已提交
270 271 272 273 274
	dput(dentry);
	exp_put(exp);
	return err;
}

275 276 277 278 279 280 281
static int nfsd_break_lease(struct inode *inode)
{
	if (!S_ISREG(inode->i_mode))
		return 0;
	return break_lease(inode, O_WRONLY | O_NONBLOCK);
}

282 283 284 285 286 287 288 289 290 291 292 293
/*
 * Commit metadata changes to stable storage.
 */
static int
commit_metadata(struct svc_fh *fhp)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	const struct export_operations *export_ops = inode->i_sb->s_export_op;

	if (!EX_ISSYNC(fhp->fh_export))
		return 0;

C
Christoph Hellwig 已提交
294 295 296
	if (export_ops->commit_metadata)
		return export_ops->commit_metadata(inode);
	return sync_inode_metadata(inode, 1);
297
}
298

L
Linus Torvalds 已提交
299 300 301 302
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
303
__be32
L
Linus Torvalds 已提交
304 305 306 307 308
nfsd_setattr(struct svc_rqst *rqstp, struct svc_fh *fhp, struct iattr *iap,
	     int check_guard, time_t guardtime)
{
	struct dentry	*dentry;
	struct inode	*inode;
M
Miklos Szeredi 已提交
309
	int		accmode = NFSD_MAY_SATTR;
A
Al Viro 已提交
310
	umode_t		ftype = 0;
311 312
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
313 314 315
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
316
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
317 318 319 320 321
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
322
	if (err)
L
Linus Torvalds 已提交
323 324 325 326 327
		goto out;

	dentry = fhp->fh_dentry;
	inode = dentry->d_inode;

328 329 330 331 332 333 334
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

335 336
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
337 338 339 340 341 342 343 344 345 346 347 348
	 * which only requires access, and "set-[ac]time-to-X" which
	 * requires ownership.
	 * So if it looks like it might be "set both to the same time which
	 * is close to now", and if inode_change_ok fails, then we
	 * convert to "set to now" instead of "set to explicit time"
	 *
	 * We only call inode_change_ok as the last test as technically
	 * it is not an interface that we should be using.  It is only
	 * valid if the filesystem does not define it's own i_op->setattr.
	 */
#define BOTH_TIME_SET (ATTR_ATIME_SET | ATTR_MTIME_SET)
#define	MAX_TOUCH_TIME_ERROR (30*60)
349 350 351 352 353 354 355 356
	if ((iap->ia_valid & BOTH_TIME_SET) == BOTH_TIME_SET &&
	    iap->ia_mtime.tv_sec == iap->ia_atime.tv_sec) {
		/*
		 * Looks probable.
		 *
		 * Now just make sure time is in the right ballpark.
		 * Solaris, at least, doesn't seem to care what the time
		 * request is.  We require it be within 30 minutes of now.
L
Linus Torvalds 已提交
357
		 */
358 359 360 361 362 363 364 365 366 367 368 369
		time_t delta = iap->ia_atime.tv_sec - get_seconds();
		if (delta < 0)
			delta = -delta;
		if (delta < MAX_TOUCH_TIME_ERROR &&
		    inode_change_ok(inode, iap) != 0) {
			/*
			 * Turn off ATTR_[AM]TIME_SET but leave ATTR_[AM]TIME.
			 * This will cause notify_change to set these times
			 * to "now"
			 */
			iap->ia_valid &= ~BOTH_TIME_SET;
		}
L
Linus Torvalds 已提交
370 371
	}
	    
372 373 374 375
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
376 377
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
378 379
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
380 381 382 383
			if (err)
				goto out;
		}

384 385
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
386 387 388
			goto out_nfserr;

		size_change = 1;
389 390
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
391 392 393 394 395
			put_write_access(inode);
			goto out_nfserr;
		}
	}

396
	/* sanitize the mode change */
L
Linus Torvalds 已提交
397 398
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
399
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
400 401 402
	}

	/* Revoke setuid/setgid on chown */
403 404 405
	if (!S_ISDIR(inode->i_mode) &&
	    (((iap->ia_valid & ATTR_UID) && iap->ia_uid != inode->i_uid) ||
	     ((iap->ia_valid & ATTR_GID) && iap->ia_gid != inode->i_gid))) {
406 407 408
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
409
			iap->ia_mode &= ~S_ISUID;
410 411 412 413 414
			if (iap->ia_mode & S_IXGRP)
				iap->ia_mode &= ~S_ISGID;
		} else {
			/* set ATTR_KILL_* bits and let VFS handle it */
			iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
415
		}
L
Linus Torvalds 已提交
416 417 418 419 420 421 422 423
	}

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
424
		host_err = nfsd_break_lease(inode);
425 426
		if (host_err)
			goto out_nfserr;
L
Linus Torvalds 已提交
427
		fh_lock(fhp);
428

429 430
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
431 432 433 434 435
		fh_unlock(fhp);
	}
	if (size_change)
		put_write_access(inode);
	if (!err)
436
		commit_metadata(fhp);
L
Linus Torvalds 已提交
437 438 439 440
out:
	return err;

out_nfserr:
441
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
442 443 444
	goto out;
}

445 446 447 448 449 450
#if defined(CONFIG_NFSD_V2_ACL) || \
    defined(CONFIG_NFSD_V3_ACL) || \
    defined(CONFIG_NFSD_V4)
static ssize_t nfsd_getxattr(struct dentry *dentry, char *key, void **buf)
{
	ssize_t buflen;
451
	ssize_t ret;
452 453 454 455

	buflen = vfs_getxattr(dentry, key, NULL, 0);
	if (buflen <= 0)
		return buflen;
L
Linus Torvalds 已提交
456

457 458 459 460
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

461 462 463 464
	ret = vfs_getxattr(dentry, key, *buf, buflen);
	if (ret < 0)
		kfree(*buf);
	return ret;
465 466 467 468
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
static int
set_nfsv4_acl_one(struct dentry *dentry, struct posix_acl *pacl, char *key)
{
	int len;
	size_t buflen;
	char *buf = NULL;
	int error = 0;

	buflen = posix_acl_xattr_size(pacl->a_count);
	buf = kmalloc(buflen, GFP_KERNEL);
	error = -ENOMEM;
	if (buf == NULL)
		goto out;

	len = posix_acl_to_xattr(pacl, buf, buflen);
	if (len < 0) {
		error = len;
		goto out;
	}

489
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
490 491 492 493 494
out:
	kfree(buf);
	return error;
}

495
__be32
L
Linus Torvalds 已提交
496 497 498
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
499 500
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
501 502 503 504 505 506
	struct dentry *dentry;
	struct inode *inode;
	struct posix_acl *pacl = NULL, *dpacl = NULL;
	unsigned int flags = 0;

	/* Get inode */
507
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
508
	if (error)
509
		return error;
L
Linus Torvalds 已提交
510 511 512 513 514 515

	dentry = fhp->fh_dentry;
	inode = dentry->d_inode;
	if (S_ISDIR(inode->i_mode))
		flags = NFS4_ACL_DIR;

516 517
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
518
		return nfserr_attrnotsupp;
519
	} else if (host_error < 0)
L
Linus Torvalds 已提交
520 521
		goto out_nfserr;

522 523
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
524
		goto out_release;
L
Linus Torvalds 已提交
525

526
	if (S_ISDIR(inode->i_mode))
527
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
528

529
out_release:
L
Linus Torvalds 已提交
530 531 532
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
533
	if (host_error == -EOPNOTSUPP)
534
		return nfserr_attrnotsupp;
535
	else
536
		return nfserrno(host_error);
L
Linus Torvalds 已提交
537 538 539 540 541
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
542
	void *buf = NULL;
L
Linus Torvalds 已提交
543
	struct posix_acl *pacl = NULL;
544
	int buflen;
L
Linus Torvalds 已提交
545

546 547 548 549 550
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
551 552 553 554 555 556 557 558 559 560 561 562 563 564

	pacl = posix_acl_from_xattr(buf, buflen);
	kfree(buf);
	return pacl;
}

int
nfsd4_get_nfs4_acl(struct svc_rqst *rqstp, struct dentry *dentry, struct nfs4_acl **acl)
{
	struct inode *inode = dentry->d_inode;
	int error = 0;
	struct posix_acl *pacl = NULL, *dpacl = NULL;
	unsigned int flags = 0;

565
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
566 567 568 569 570 571 572 573 574
	if (IS_ERR(pacl) && PTR_ERR(pacl) == -ENODATA)
		pacl = posix_acl_from_mode(inode->i_mode, GFP_KERNEL);
	if (IS_ERR(pacl)) {
		error = PTR_ERR(pacl);
		pacl = NULL;
		goto out;
	}

	if (S_ISDIR(inode->i_mode)) {
575
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
		if (IS_ERR(dpacl) && PTR_ERR(dpacl) == -ENODATA)
			dpacl = NULL;
		else if (IS_ERR(dpacl)) {
			error = PTR_ERR(dpacl);
			dpacl = NULL;
			goto out;
		}
		flags = NFS4_ACL_DIR;
	}

	*acl = nfs4_acl_posix_to_nfsv4(pacl, dpacl, flags);
	if (IS_ERR(*acl)) {
		error = PTR_ERR(*acl);
		*acl = NULL;
	}
 out:
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
	return error;
}

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612
#define NFSD_XATTR_JUNCTION_PREFIX XATTR_TRUSTED_PREFIX "junction."
#define NFSD_XATTR_JUNCTION_TYPE NFSD_XATTR_JUNCTION_PREFIX "type"
int nfsd4_is_junction(struct dentry *dentry)
{
	struct inode *inode = dentry->d_inode;

	if (inode == NULL)
		return 0;
	if (inode->i_mode & S_IXUGO)
		return 0;
	if (!(inode->i_mode & S_ISVTX))
		return 0;
	if (vfs_getxattr(dentry, NFSD_XATTR_JUNCTION_TYPE, NULL, 0) <= 0)
		return 0;
	return 1;
}
J
J. Bruce Fields 已提交
613
#endif /* defined(CONFIG_NFSD_V4) */
L
Linus Torvalds 已提交
614 615 616 617 618 619 620 621 622 623

#ifdef CONFIG_NFSD_V3
/*
 * Check server access rights to a file system object
 */
struct accessmap {
	u32		access;
	int		how;
};
static struct accessmap	nfs3_regaccess[] = {
M
Miklos Szeredi 已提交
624 625 626 627
    {	NFS3_ACCESS_READ,	NFSD_MAY_READ			},
    {	NFS3_ACCESS_EXECUTE,	NFSD_MAY_EXEC			},
    {	NFS3_ACCESS_MODIFY,	NFSD_MAY_WRITE|NFSD_MAY_TRUNC	},
    {	NFS3_ACCESS_EXTEND,	NFSD_MAY_WRITE			},
L
Linus Torvalds 已提交
628 629 630 631 632

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
633 634 635 636 637
    {	NFS3_ACCESS_READ,	NFSD_MAY_READ			},
    {	NFS3_ACCESS_LOOKUP,	NFSD_MAY_EXEC			},
    {	NFS3_ACCESS_MODIFY,	NFSD_MAY_EXEC|NFSD_MAY_WRITE|NFSD_MAY_TRUNC},
    {	NFS3_ACCESS_EXTEND,	NFSD_MAY_EXEC|NFSD_MAY_WRITE	},
    {	NFS3_ACCESS_DELETE,	NFSD_MAY_REMOVE			},
L
Linus Torvalds 已提交
638 639 640 641 642 643 644 645 646 647 648 649

    {	0,			0				}
};

static struct accessmap	nfs3_anyaccess[] = {
	/* Some clients - Solaris 2.6 at least, make an access call
	 * to the server to check for access for things like /dev/null
	 * (which really, the server doesn't care about).  So
	 * We provide simple access checking for them, looking
	 * mainly at mode bits, and we make sure to ignore read-only
	 * filesystem checks
	 */
M
Miklos Szeredi 已提交
650 651 652 653
    {	NFS3_ACCESS_READ,	NFSD_MAY_READ			},
    {	NFS3_ACCESS_EXECUTE,	NFSD_MAY_EXEC			},
    {	NFS3_ACCESS_MODIFY,	NFSD_MAY_WRITE|NFSD_MAY_LOCAL_ACCESS	},
    {	NFS3_ACCESS_EXTEND,	NFSD_MAY_WRITE|NFSD_MAY_LOCAL_ACCESS	},
L
Linus Torvalds 已提交
654 655 656 657

    {	0,			0				}
};

658
__be32
L
Linus Torvalds 已提交
659 660 661 662 663 664
nfsd_access(struct svc_rqst *rqstp, struct svc_fh *fhp, u32 *access, u32 *supported)
{
	struct accessmap	*map;
	struct svc_export	*export;
	struct dentry		*dentry;
	u32			query, result = 0, sresult = 0;
665
	__be32			error;
L
Linus Torvalds 已提交
666

M
Miklos Szeredi 已提交
667
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684
	if (error)
		goto out;

	export = fhp->fh_export;
	dentry = fhp->fh_dentry;

	if (S_ISREG(dentry->d_inode->i_mode))
		map = nfs3_regaccess;
	else if (S_ISDIR(dentry->d_inode->i_mode))
		map = nfs3_diraccess;
	else
		map = nfs3_anyaccess;


	query = *access;
	for  (; map->access; map++) {
		if (map->access & query) {
685
			__be32 err2;
L
Linus Torvalds 已提交
686 687 688

			sresult |= map->access;

689
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
			switch (err2) {
			case nfs_ok:
				result |= map->access;
				break;
				
			/* the following error codes just mean the access was not allowed,
			 * rather than an error occurred */
			case nfserr_rofs:
			case nfserr_acces:
			case nfserr_perm:
				/* simply don't "or" in the access bit. */
				break;
			default:
				error = err2;
				goto out;
			}
		}
	}
	*access = result;
	if (supported)
		*supported = sresult;

 out:
	return error;
}
#endif /* CONFIG_NFSD_V3 */

717 718 719
static int nfsd_open_break_lease(struct inode *inode, int access)
{
	unsigned int mode;
L
Linus Torvalds 已提交
720

721 722 723 724 725
	if (access & NFSD_MAY_NOT_BREAK_LEASE)
		return 0;
	mode = (access & NFSD_MAY_WRITE) ? O_WRONLY : O_RDONLY;
	return break_lease(inode, mode | O_NONBLOCK);
}
L
Linus Torvalds 已提交
726 727 728 729 730 731

/*
 * Open an existing file or directory.
 * The access argument indicates the type of open (read/write/lock)
 * N.B. After this call fhp needs an fh_put
 */
732
__be32
A
Al Viro 已提交
733
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
L
Linus Torvalds 已提交
734 735 736 737
			int access, struct file **filp)
{
	struct dentry	*dentry;
	struct inode	*inode;
738 739
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
740
	int		host_err = 0;
L
Linus Torvalds 已提交
741

742 743
	validate_process_creds();

L
Linus Torvalds 已提交
744 745 746 747 748
	/*
	 * If we get here, then the client has already done an "open",
	 * and (hopefully) checked permission - so allow OWNER_OVERRIDE
	 * in case a chmod has now revoked permission.
	 */
M
Miklos Szeredi 已提交
749
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
750 751 752 753 754 755 756 757 758 759
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
	inode = dentry->d_inode;

	/* Disallow write access to files with the append-only bit set
	 * or any access when mandatory locking enabled
	 */
	err = nfserr_perm;
M
Miklos Szeredi 已提交
760
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
761
		goto out;
J
J. Bruce Fields 已提交
762 763 764 765 766 767
	/*
	 * We must ignore files (but only files) which might have mandatory
	 * locks on them because there is no way to know if the accesser has
	 * the lock.
	 */
	if (S_ISREG((inode)->i_mode) && mandatory_lock(inode))
L
Linus Torvalds 已提交
768 769 770 771 772
		goto out;

	if (!inode->i_fop)
		goto out;

773
	host_err = nfsd_open_break_lease(inode, access);
774
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
775 776
		goto out_nfserr;

M
Miklos Szeredi 已提交
777 778
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
779 780 781
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
782
	}
783
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
784
			    flags, current_cred());
L
Linus Torvalds 已提交
785
	if (IS_ERR(*filp))
786
		host_err = PTR_ERR(*filp);
787 788
	else
		host_err = ima_file_check(*filp, access);
L
Linus Torvalds 已提交
789
out_nfserr:
790
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
791
out:
792
	validate_process_creds();
L
Linus Torvalds 已提交
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814
	return err;
}

/*
 * Close a file.
 */
void
nfsd_close(struct file *filp)
{
	fput(filp);
}

/*
 * Obtain the readahead parameters for the file
 * specified by (dev, ino).
 */

static inline struct raparms *
nfsd_get_raparms(dev_t dev, ino_t ino)
{
	struct raparms	*ra, **rap, **frap = NULL;
	int depth = 0;
815 816 817 818 819
	unsigned int hash;
	struct raparm_hbucket *rab;

	hash = jhash_2words(dev, ino, 0xfeedbeef) & RAPARM_HASH_MASK;
	rab = &raparm_hash[hash];
L
Linus Torvalds 已提交
820

821 822
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
823 824 825 826 827 828
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
829
	depth = nfsdstats.ra_size;
L
Linus Torvalds 已提交
830
	if (!frap) {	
831
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
832 833 834 835 836 837 838
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
839
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
840
found:
841
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
842
		*rap = ra->p_next;
843 844
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
845 846 847
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
848
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
849 850 851 852
	return ra;
}

/*
853 854 855
 * Grab and keep cached pages associated with a file in the svc_rqst
 * so that they can be passed to the network sendmsg/sendpage routines
 * directly. They will be released after the sending has completed.
L
Linus Torvalds 已提交
856 857
 */
static int
858 859
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
860
{
861
	struct svc_rqst *rqstp = sd->u.data;
862
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
863 864 865 866
	struct page *page = buf->page;
	size_t size;

	size = sd->len;
L
Linus Torvalds 已提交
867 868 869

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
870 871 872
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
873
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
874
		rqstp->rq_res.page_len = size;
875
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
876
		get_page(page);
877 878
		if (*pp)
			put_page(*pp);
879 880
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
881
		rqstp->rq_res.page_len += size;
882
	} else
L
Linus Torvalds 已提交
883 884 885 886 887
		rqstp->rq_res.page_len += size;

	return size;
}

888 889 890 891 892 893
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

894
static __be32
L
Linus Torvalds 已提交
895 896 897 898
nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
              loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
{
	mm_segment_t	oldfs;
899 900
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
901 902

	err = nfserr_perm;
903

904 905 906 907 908 909 910 911
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

912
		rqstp->rq_resused = 1;
913
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
914 915 916
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
917
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
918 919 920
		set_fs(oldfs);
	}

921 922 923
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
924
		err = 0;
925
		fsnotify_access(file);
L
Linus Torvalds 已提交
926
	} else 
927
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
928 929 930
	return err;
}

931 932 933
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
934
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
935

936
	mutex_lock(&dentry->d_inode->i_mutex);
937
	notify_change(dentry, &ia);
938
	mutex_unlock(&dentry->d_inode->i_mutex);
939 940
}

941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
/*
 * Gathered writes: If another process is currently writing to the file,
 * there's a high chance this is another nfsd (triggered by a bulk write
 * from a client's biod). Rather than syncing the file with each write
 * request, we sleep for 10 msec.
 *
 * I don't know if this roughly approximates C. Juszak's idea of
 * gathered writes, but it's a nice and simple solution (IMHO), and it
 * seems to work:-)
 *
 * Note: we do this only in the NFSv2 case, since v3 and higher have a
 * better tool (separate unstable writes and commits) for solving this
 * problem.
 */
static int wait_for_concurrent_writes(struct file *file)
{
	struct inode *inode = file->f_path.dentry->d_inode;
	static ino_t last_ino;
	static dev_t last_dev;
	int err = 0;

	if (atomic_read(&inode->i_writecount) > 1
	    || (last_ino == inode->i_ino && last_dev == inode->i_sb->s_dev)) {
		dprintk("nfsd: write defer %d\n", task_pid_nr(current));
		msleep(10);
		dprintk("nfsd: write resume %d\n", task_pid_nr(current));
	}

	if (inode->i_state & I_DIRTY) {
		dprintk("nfsd: write sync %d\n", task_pid_nr(current));
971
		err = vfs_fsync(file, 0);
972 973 974 975 976 977
	}
	last_ino = inode->i_ino;
	last_dev = inode->i_sb->s_dev;
	return err;
}

978
static __be32
L
Linus Torvalds 已提交
979 980
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
981
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
982 983 984 985 986
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
987 988
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
989
	int			stable = *stablep;
990
	int			use_wgather;
L
Linus Torvalds 已提交
991

992
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
993 994 995 996 997 998 999 1000
	inode = dentry->d_inode;
	exp   = fhp->fh_export;

	/*
	 * Request sync writes if
	 *  -	the sync export option has been set, or
	 *  -	the client requested O_SYNC behavior (NFSv3 feature).
	 *  -   The file system doesn't support fsync().
1001
	 * When NFSv2 gathered writes have been configured for this volume,
L
Linus Torvalds 已提交
1002 1003
	 * flushing the data to disk is handled separately below.
	 */
1004
	use_wgather = (rqstp->rq_vers == 2) && EX_WGATHER(exp);
L
Linus Torvalds 已提交
1005

1006
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
1007 1008 1009 1010 1011 1012
	       stable = 2;
	       *stablep = 2; /* FILE_SYNC */
	}

	if (!EX_ISSYNC(exp))
		stable = 0;
1013
	if (stable && !use_wgather) {
J
Jonathan Corbet 已提交
1014
		spin_lock(&file->f_lock);
L
Linus Torvalds 已提交
1015
		file->f_flags |= O_SYNC;
J
Jonathan Corbet 已提交
1016 1017
		spin_unlock(&file->f_lock);
	}
L
Linus Torvalds 已提交
1018 1019 1020

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
1021
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1022
	set_fs(oldfs);
1023 1024 1025 1026
	if (host_err < 0)
		goto out_nfserr;
	*cnt = host_err;
	nfsdstats.io_write += host_err;
1027
	fsnotify_modify(file);
L
Linus Torvalds 已提交
1028 1029

	/* clear setuid/setgid flag after write */
1030
	if (inode->i_mode & (S_ISUID | S_ISGID))
1031
		kill_suid(dentry);
L
Linus Torvalds 已提交
1032

1033
	if (stable && use_wgather)
1034
		host_err = wait_for_concurrent_writes(file);
L
Linus Torvalds 已提交
1035

1036
out_nfserr:
1037
	dprintk("nfsd: write complete host_err=%d\n", host_err);
1038
	if (host_err >= 0)
L
Linus Torvalds 已提交
1039
		err = 0;
1040
	else
1041
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1042 1043 1044 1045 1046 1047 1048 1049
	return err;
}

/*
 * Read data from a file. count must contain the requested read count
 * on entry. On return, *count contains the number of bytes actually read.
 * N.B. After this call fhp needs an fh_put
 */
1050
__be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
	loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
{
	struct file *file;
	struct inode *inode;
	struct raparms	*ra;
	__be32 err;

	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (err)
		return err;

	inode = file->f_path.dentry->d_inode;

	/* Get readahead parameters */
	ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);

	if (ra && ra->p_set)
		file->f_ra = ra->p_ra;

	err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);

	/* Write back readahead params */
	if (ra) {
		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
		spin_lock(&rab->pb_lock);
		ra->p_ra = file->f_ra;
		ra->p_set = 1;
		ra->p_count--;
		spin_unlock(&rab->pb_lock);
	}

	nfsd_close(file);
	return err;
}

1086
/* As above, but use the provided file descriptor. */
1087
__be32
1088
nfsd_read_file(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
L
Linus Torvalds 已提交
1089 1090 1091
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1092
	__be32		err;
L
Linus Torvalds 已提交
1093 1094

	if (file) {
1095
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1096
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1097 1098 1099
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
1100 1101
	} else /* Note file may still be NULL in NFSv4 special stateid case: */
		err = nfsd_read(rqstp, fhp, offset, vec, vlen, count);
L
Linus Torvalds 已提交
1102 1103 1104 1105 1106 1107 1108 1109 1110
out:
	return err;
}

/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1111
__be32
L
Linus Torvalds 已提交
1112
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1113
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1114 1115
		int *stablep)
{
1116
	__be32			err = 0;
L
Linus Torvalds 已提交
1117 1118

	if (file) {
1119
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1120
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1121 1122 1123 1124 1125
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1126
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
		if (err)
			goto out;

		if (cnt)
			err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen,
					     cnt, stablep);
		nfsd_close(file);
	}
out:
	return err;
}

#ifdef CONFIG_NFSD_V3
/*
 * Commit all pending writes to stable storage.
1142 1143 1144
 *
 * Note: we only guarantee that data that lies within the range specified
 * by the 'offset' and 'count' parameters will be synced.
L
Linus Torvalds 已提交
1145 1146 1147 1148
 *
 * Unfortunately we cannot lock the file to make sure we return full WCC
 * data to the client, as locking happens lower down in the filesystem.
 */
1149
__be32
L
Linus Torvalds 已提交
1150 1151 1152 1153
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1154 1155
	loff_t		end = LLONG_MAX;
	__be32		err = nfserr_inval;
L
Linus Torvalds 已提交
1156

1157 1158 1159 1160 1161 1162 1163
	if (offset < 0)
		goto out;
	if (count != 0) {
		end = offset + (loff_t)count - 1;
		if (end < offset)
			goto out;
	}
L
Linus Torvalds 已提交
1164

1165 1166
	err = nfsd_open(rqstp, fhp, S_IFREG,
			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &file);
M
Miklos Szeredi 已提交
1167
	if (err)
1168
		goto out;
L
Linus Torvalds 已提交
1169
	if (EX_ISSYNC(fhp->fh_export)) {
1170
		int err2 = vfs_fsync_range(file, offset, end, 0);
1171 1172 1173 1174

		if (err2 != -EINVAL)
			err = nfserrno(err2);
		else
L
Linus Torvalds 已提交
1175 1176 1177 1178
			err = nfserr_notsupp;
	}

	nfsd_close(file);
1179
out:
L
Linus Torvalds 已提交
1180 1181 1182 1183
	return err;
}
#endif /* CONFIG_NFSD_V3 */

A
Adrian Bunk 已提交
1184
static __be32
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
nfsd_create_setattr(struct svc_rqst *rqstp, struct svc_fh *resfhp,
			struct iattr *iap)
{
	/*
	 * Mode has already been set earlier in create:
	 */
	iap->ia_valid &= ~ATTR_MODE;
	/*
	 * Setting uid/gid works only for root.  Irix appears to
	 * send along the gid on create when it tries to implement
	 * setgid directories via NFS:
	 */
1197
	if (current_fsuid() != 0)
1198 1199 1200 1201 1202 1203
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218
/* HPUX client sometimes creates a file in mode 000, and sets size to 0.
 * setting size to 0 may fail for some specific file systems by the permission
 * checking which requires WRITE permission but the mode is 000.
 * we ignore the resizing(to 0) on the just new created file, since the size is
 * 0 after file created.
 *
 * call this only after vfs_create() is called.
 * */
static void
nfsd_check_ignore_resizing(struct iattr *iap)
{
	if ((iap->ia_valid & ATTR_SIZE) && (iap->ia_size == 0))
		iap->ia_valid &= ~ATTR_SIZE;
}

L
Linus Torvalds 已提交
1219 1220 1221 1222 1223 1224 1225 1226
/*
 * Create a file (regular, directory, device, fifo); UNIX sockets 
 * not yet implemented.
 * If the response fh has been verified, the parent directory should
 * already be locked. Note that the parent directory is left locked.
 *
 * N.B. Every call to nfsd_create needs an fh_put for _both_ fhp and resfhp
 */
1227
__be32
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232 1233
nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
		char *fname, int flen, struct iattr *iap,
		int type, dev_t rdev, struct svc_fh *resfhp)
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1234
	__be32		err;
1235
	__be32		err2;
1236
	int		host_err;
L
Linus Torvalds 已提交
1237 1238 1239 1240 1241 1242 1243 1244

	err = nfserr_perm;
	if (!flen)
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;

M
Miklos Szeredi 已提交
1245
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1246 1247 1248 1249 1250 1251 1252
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	err = nfserr_notdir;
A
Al Viro 已提交
1253
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260
		goto out;
	/*
	 * Check whether the response file handle has been verified yet.
	 * If it has, the parent directory should already be locked.
	 */
	if (!resfhp->fh_dentry) {
		/* called from nfsd_proc_mkdir, or possibly nfsd3_proc_create */
P
Peter Zijlstra 已提交
1261
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1262
		dchild = lookup_one_len(fname, dentry, flen);
1263
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
		if (IS_ERR(dchild))
			goto out_nfserr;
		err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
		if (err)
			goto out;
	} else {
		/* called from nfsd_proc_create */
		dchild = dget(resfhp->fh_dentry);
		if (!fhp->fh_locked) {
			/* not actually possible */
			printk(KERN_ERR
				"nfsd_create: parent %s/%s not locked!\n",
				dentry->d_parent->d_name.name,
				dentry->d_name.name);
A
Al Viro 已提交
1278
			err = nfserr_io;
L
Linus Torvalds 已提交
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
			goto out;
		}
	}
	/*
	 * Make sure the child dentry is still negative ...
	 */
	err = nfserr_exist;
	if (dchild->d_inode) {
		dprintk("nfsd_create: dentry %s/%s not negative!\n",
			dentry->d_name.name, dchild->d_name.name);
		goto out; 
	}

	if (!(iap->ia_valid & ATTR_MODE))
		iap->ia_mode = 0;
	iap->ia_mode = (iap->ia_mode & S_IALLUGO) | type;

1296 1297 1298 1299 1300 1301 1302
	err = nfserr_inval;
	if (!S_ISREG(type) && !S_ISDIR(type) && !special_file(type)) {
		printk(KERN_WARNING "nfsd: bad file type %o in nfsd_create\n",
		       type);
		goto out;
	}

A
Al Viro 已提交
1303
	host_err = fh_want_write(fhp);
1304 1305 1306
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1307 1308 1309
	/*
	 * Get the dir op function pointer.
	 */
1310
	err = 0;
L
Linus Torvalds 已提交
1311 1312
	switch (type) {
	case S_IFREG:
1313
		host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1314 1315
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
L
Linus Torvalds 已提交
1316 1317
		break;
	case S_IFDIR:
1318
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1319 1320 1321 1322 1323
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1324
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1325 1326
		break;
	}
1327
	if (host_err < 0) {
A
Al Viro 已提交
1328
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1329
		goto out_nfserr;
1330
	}
L
Linus Torvalds 已提交
1331

1332
	err = nfsd_create_setattr(rqstp, resfhp, iap);
L
Linus Torvalds 已提交
1333

1334 1335 1336 1337 1338 1339
	/*
	 * nfsd_setattr already committed the child.  Transactional filesystems
	 * had a chance to commit changes for both parent and child
	 * simultaneously making the following commit_metadata a noop.
	 */
	err2 = nfserrno(commit_metadata(fhp));
1340 1341
	if (err2)
		err = err2;
A
Al Viro 已提交
1342
	fh_drop_write(fhp);
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
	/*
	 * Update the file handle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
out:
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
	return err;

out_nfserr:
1354
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1355 1356 1357 1358
	goto out;
}

#ifdef CONFIG_NFSD_V3
1359 1360 1361 1362 1363 1364 1365

static inline int nfsd_create_is_exclusive(int createmode)
{
	return createmode == NFS3_CREATE_EXCLUSIVE
	       || createmode == NFS4_CREATE_EXCLUSIVE4_1;
}

L
Linus Torvalds 已提交
1366
/*
1367
 * NFSv3 and NFSv4 version of nfsd_create
L
Linus Torvalds 已提交
1368
 */
1369
__be32
1370
do_nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
L
Linus Torvalds 已提交
1371 1372
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1373
	        bool *truncp, bool *created)
L
Linus Torvalds 已提交
1374 1375 1376
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1377 1378
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388
	__u32		v_mtime=0, v_atime=0;

	err = nfserr_perm;
	if (!flen)
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;
	if (!(iap->ia_valid & ATTR_MODE))
		iap->ia_mode = 0;
1389
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
1390 1391 1392 1393 1394 1395 1396 1397 1398
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	/* Get all the sanity checks out of the way before
	 * we lock the parent. */
	err = nfserr_notdir;
A
Al Viro 已提交
1399
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1400
		goto out;
P
Peter Zijlstra 已提交
1401
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1402 1403 1404 1405 1406

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1407
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1408 1409 1410
	if (IS_ERR(dchild))
		goto out_nfserr;

1411 1412 1413 1414 1415 1416 1417
	/* If file doesn't exist, check for permissions to create one */
	if (!dchild->d_inode) {
		err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
		if (err)
			goto out;
	}

L
Linus Torvalds 已提交
1418 1419 1420 1421
	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
	if (err)
		goto out;

1422
	if (nfsd_create_is_exclusive(createmode)) {
1423
		/* solaris7 gets confused (bugid 4218508) if these have
1424 1425 1426 1427
		 * the high bit set, so just clear the high bits. If this is
		 * ever changed to use different attrs for storing the
		 * verifier, then do_open_lookup() will also need to be fixed
		 * accordingly.
L
Linus Torvalds 已提交
1428 1429 1430 1431 1432
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
A
Al Viro 已提交
1433
	host_err = fh_want_write(fhp);
1434 1435
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462
	if (dchild->d_inode) {
		err = 0;

		switch (createmode) {
		case NFS3_CREATE_UNCHECKED:
			if (! S_ISREG(dchild->d_inode->i_mode))
				err = nfserr_exist;
			else if (truncp) {
				/* in nfsv4, we need to treat this case a little
				 * differently.  we don't want to truncate the
				 * file now; this would be wrong if the OPEN
				 * fails for some other reason.  furthermore,
				 * if the size is nonzero, we should ignore it
				 * according to spec!
				 */
				*truncp = (iap->ia_valid & ATTR_SIZE) && !iap->ia_size;
			}
			else {
				iap->ia_valid &= ATTR_SIZE;
				goto set_attr;
			}
			break;
		case NFS3_CREATE_EXCLUSIVE:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
			    && dchild->d_inode->i_size  == 0 )
				break;
1463 1464 1465 1466 1467
		case NFS4_CREATE_EXCLUSIVE4_1:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
			    && dchild->d_inode->i_size  == 0 )
				goto set_attr;
L
Linus Torvalds 已提交
1468 1469 1470 1471
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
A
Al Viro 已提交
1472
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1473 1474 1475
		goto out;
	}

1476
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1477
	if (host_err < 0) {
A
Al Viro 已提交
1478
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1479
		goto out_nfserr;
1480
	}
1481 1482
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1483

1484 1485
	nfsd_check_ignore_resizing(iap);

1486
	if (nfsd_create_is_exclusive(createmode)) {
1487
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1488
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1489
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1490 1491 1492 1493 1494 1495 1496 1497
		/* XXX someone who knows this better please fix it for nsec */ 
		iap->ia_mtime.tv_sec = v_mtime;
		iap->ia_atime.tv_sec = v_atime;
		iap->ia_mtime.tv_nsec = 0;
		iap->ia_atime.tv_nsec = 0;
	}

 set_attr:
1498 1499 1500 1501 1502 1503 1504
	err = nfsd_create_setattr(rqstp, resfhp, iap);

	/*
	 * nfsd_setattr already committed the child (and possibly also the parent).
	 */
	if (!err)
		err = nfserrno(commit_metadata(fhp));
1505

A
Al Viro 已提交
1506
	fh_drop_write(fhp);
1507 1508 1509 1510 1511
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1512 1513 1514 1515 1516 1517 1518 1519

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1520
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1521 1522 1523 1524 1525 1526 1527 1528 1529
	goto out;
}
#endif /* CONFIG_NFSD_V3 */

/*
 * Read a symlink. On entry, *lenp must contain the maximum path length that
 * fits into the buffer. On return, it contains the true length.
 * N.B. After this call fhp needs an fh_put
 */
1530
__be32
L
Linus Torvalds 已提交
1531 1532 1533 1534 1535
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1536 1537
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1538

M
Miklos Szeredi 已提交
1539
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1540 1541 1542 1543 1544 1545 1546
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
	inode = dentry->d_inode;

	err = nfserr_inval;
A
Al Viro 已提交
1547
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1548 1549
		goto out;

1550
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1551 1552 1553 1554 1555
	/* N.B. Why does this call need a get_fs()??
	 * Remove the set_fs and watch the fireworks:-) --okir
	 */

	oldfs = get_fs(); set_fs(KERNEL_DS);
1556
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1557 1558
	set_fs(oldfs);

1559
	if (host_err < 0)
L
Linus Torvalds 已提交
1560
		goto out_nfserr;
1561
	*lenp = host_err;
L
Linus Torvalds 已提交
1562 1563 1564 1565 1566
	err = 0;
out:
	return err;

out_nfserr:
1567
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1568 1569 1570 1571 1572 1573 1574
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1575
__be32
L
Linus Torvalds 已提交
1576 1577 1578 1579 1580 1581 1582
nfsd_symlink(struct svc_rqst *rqstp, struct svc_fh *fhp,
				char *fname, int flen,
				char *path,  int plen,
				struct svc_fh *resfhp,
				struct iattr *iap)
{
	struct dentry	*dentry, *dnew;
1583 1584
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1585 1586 1587 1588 1589 1590 1591 1592

	err = nfserr_noent;
	if (!flen || !plen)
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;

M
Miklos Szeredi 已提交
1593
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1594 1595 1596 1597 1598
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1599
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1600 1601 1602
	if (IS_ERR(dnew))
		goto out_nfserr;

A
Al Viro 已提交
1603
	host_err = fh_want_write(fhp);
1604 1605 1606
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1607 1608 1609
	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1610
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1611 1612 1613
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1614
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1615 1616 1617
			kfree(path_alloced);
		}
	} else
1618
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
1619
	err = nfserrno(host_err);
1620 1621
	if (!err)
		err = nfserrno(commit_metadata(fhp));
L
Linus Torvalds 已提交
1622 1623
	fh_unlock(fhp);

A
Al Viro 已提交
1624
	fh_drop_write(fhp);
1625

L
Linus Torvalds 已提交
1626 1627 1628 1629 1630 1631 1632
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1633
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639 1640
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1641
__be32
L
Linus Torvalds 已提交
1642 1643 1644 1645
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
1646
	struct inode	*dirp;
1647 1648
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1649

M
Miklos Szeredi 已提交
1650
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1651 1652
	if (err)
		goto out;
1653
	err = fh_verify(rqstp, tfhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1654 1655
	if (err)
		goto out;
1656 1657 1658
	err = nfserr_isdir;
	if (S_ISDIR(tfhp->fh_dentry->d_inode->i_mode))
		goto out;
L
Linus Torvalds 已提交
1659 1660 1661 1662 1663 1664 1665
	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

P
Peter Zijlstra 已提交
1666
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1667 1668 1669 1670
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1671
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1672 1673 1674 1675 1676
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;

A
Al Viro 已提交
1677
	host_err = fh_want_write(tfhp);
1678 1679 1680 1681
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1682 1683 1684 1685
	err = nfserr_noent;
	if (!dold->d_inode)
		goto out_drop_write;
	host_err = nfsd_break_lease(dold->d_inode);
1686 1687
	if (host_err) {
		err = nfserrno(host_err);
1688
		goto out_drop_write;
1689
	}
1690 1691
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
1692 1693 1694
		err = nfserrno(commit_metadata(ffhp));
		if (!err)
			err = nfserrno(commit_metadata(tfhp));
L
Linus Torvalds 已提交
1695
	} else {
1696
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1697 1698
			err = nfserr_acces;
		else
1699
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1700
	}
1701
out_drop_write:
A
Al Viro 已提交
1702
	fh_drop_write(tfhp);
1703
out_dput:
L
Linus Torvalds 已提交
1704
	dput(dnew);
1705 1706
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1707 1708 1709 1710
out:
	return err;

out_nfserr:
1711
	err = nfserrno(host_err);
1712
	goto out_unlock;
L
Linus Torvalds 已提交
1713 1714 1715 1716 1717 1718
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1719
__be32
L
Linus Torvalds 已提交
1720 1721 1722 1723 1724
nfsd_rename(struct svc_rqst *rqstp, struct svc_fh *ffhp, char *fname, int flen,
			    struct svc_fh *tfhp, char *tname, int tlen)
{
	struct dentry	*fdentry, *tdentry, *odentry, *ndentry, *trap;
	struct inode	*fdir, *tdir;
1725 1726
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1727

M
Miklos Szeredi 已提交
1728
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1729 1730
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1731
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
	if (err)
		goto out;

	fdentry = ffhp->fh_dentry;
	fdir = fdentry->d_inode;

	tdentry = tfhp->fh_dentry;
	tdir = tdentry->d_inode;

	err = (rqstp->rq_vers == 2) ? nfserr_acces : nfserr_xdev;
1742
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756
		goto out;

	err = nfserr_perm;
	if (!flen || isdotent(fname, flen) || !tlen || isdotent(tname, tlen))
		goto out;

	/* cannot use fh_lock as we need deadlock protective ordering
	 * so do it by hand */
	trap = lock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 1;
	fill_pre_wcc(ffhp);
	fill_pre_wcc(tfhp);

	odentry = lookup_one_len(fname, fdentry, flen);
1757
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1758 1759 1760
	if (IS_ERR(odentry))
		goto out_nfserr;

1761
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1762 1763
	if (!odentry->d_inode)
		goto out_dput_old;
1764
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1765 1766 1767 1768
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1769
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1770 1771
	if (IS_ERR(ndentry))
		goto out_dput_old;
1772
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1773 1774 1775
	if (ndentry == trap)
		goto out_dput_new;

1776 1777 1778
	host_err = -EXDEV;
	if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
		goto out_dput_new;
A
Al Viro 已提交
1779
	host_err = fh_want_write(ffhp);
1780 1781 1782
	if (host_err)
		goto out_dput_new;

1783
	host_err = nfsd_break_lease(odentry->d_inode);
1784 1785 1786 1787 1788 1789 1790
	if (host_err)
		goto out_drop_write;
	if (ndentry->d_inode) {
		host_err = nfsd_break_lease(ndentry->d_inode);
		if (host_err)
			goto out_drop_write;
	}
1791
	host_err = vfs_rename(fdir, odentry, tdir, ndentry);
1792 1793
	if (!host_err) {
		host_err = commit_metadata(tfhp);
1794
		if (!host_err)
1795
			host_err = commit_metadata(ffhp);
L
Linus Torvalds 已提交
1796
	}
1797
out_drop_write:
A
Al Viro 已提交
1798
	fh_drop_write(ffhp);
L
Linus Torvalds 已提交
1799 1800 1801 1802 1803
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1804
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822

	/* we cannot reply on fh_unlock on the two filehandles,
	 * as that would do the wrong thing if the two directories
	 * were the same, so again we do it by hand
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
1823
__be32
L
Linus Torvalds 已提交
1824 1825 1826 1827 1828
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1829 1830
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1831 1832 1833 1834

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1835
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1836 1837 1838
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1839
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1840 1841 1842 1843
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1844
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856
	if (IS_ERR(rdentry))
		goto out_nfserr;

	if (!rdentry->d_inode) {
		dput(rdentry);
		err = nfserr_noent;
		goto out;
	}

	if (!type)
		type = rdentry->d_inode->i_mode & S_IFMT;

A
Al Viro 已提交
1857
	host_err = fh_want_write(fhp);
1858
	if (host_err)
1859
		goto out_put;
1860

1861 1862
	host_err = nfsd_break_lease(rdentry->d_inode);
	if (host_err)
1863
		goto out_drop_write;
1864
	if (type != S_IFDIR)
1865
		host_err = vfs_unlink(dirp, rdentry);
1866
	else
1867
		host_err = vfs_rmdir(dirp, rdentry);
1868 1869
	if (!host_err)
		host_err = commit_metadata(fhp);
1870
out_drop_write:
A
Al Viro 已提交
1871
	fh_drop_write(fhp);
1872 1873 1874
out_put:
	dput(rdentry);

L
Linus Torvalds 已提交
1875
out_nfserr:
1876
	err = nfserrno(host_err);
1877 1878
out:
	return err;
L
Linus Torvalds 已提交
1879 1880
}

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
/*
 * We do this buffering because we must not call back into the file
 * system's ->lookup() method from the filldir callback. That may well
 * deadlock a number of file systems.
 *
 * This is based heavily on the implementation of same in XFS.
 */
struct buffered_dirent {
	u64		ino;
	loff_t		offset;
	int		namlen;
	unsigned int	d_type;
	char		name[];
};

struct readdir_data {
	char		*dirent;
	size_t		used;
1899
	int		full;
1900 1901 1902 1903 1904 1905 1906 1907 1908 1909
};

static int nfsd_buffered_filldir(void *__buf, const char *name, int namlen,
				 loff_t offset, u64 ino, unsigned int d_type)
{
	struct readdir_data *buf = __buf;
	struct buffered_dirent *de = (void *)(buf->dirent + buf->used);
	unsigned int reclen;

	reclen = ALIGN(sizeof(struct buffered_dirent) + namlen, sizeof(u64));
1910 1911
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1912
		return -EINVAL;
1913
	}
1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924

	de->namlen = namlen;
	de->offset = offset;
	de->ino = ino;
	de->d_type = d_type;
	memcpy(de->name, name, namlen);
	buf->used += reclen;

	return 0;
}

1925 1926
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1927
{
1928 1929
	struct readdir_data buf;
	struct buffered_dirent *de;
1930
	int host_err;
1931 1932
	int size;
	loff_t offset;
1933

1934 1935
	buf.dirent = (void *)__get_free_page(GFP_KERNEL);
	if (!buf.dirent)
1936
		return nfserrno(-ENOMEM);
1937 1938

	offset = *offsetp;
1939

1940
	while (1) {
1941
		struct inode *dir_inode = file->f_path.dentry->d_inode;
1942 1943
		unsigned int reclen;

1944
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1945
		buf.used = 0;
1946
		buf.full = 0;
1947 1948

		host_err = vfs_readdir(file, nfsd_buffered_filldir, &buf);
1949 1950 1951 1952
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1953 1954 1955 1956 1957 1958 1959
			break;

		size = buf.used;

		if (!size)
			break;

1960 1961 1962 1963 1964 1965 1966 1967 1968
		/*
		 * Various filldir functions may end up calling back into
		 * lookup_one_len() and the file system's ->lookup() method.
		 * These expect i_mutex to be held, as it would within readdir.
		 */
		host_err = mutex_lock_killable(&dir_inode->i_mutex);
		if (host_err)
			break;

1969 1970 1971 1972 1973 1974
		de = (struct buffered_dirent *)buf.dirent;
		while (size > 0) {
			offset = de->offset;

			if (func(cdp, de->name, de->namlen, de->offset,
				 de->ino, de->d_type))
1975
				break;
1976 1977

			if (cdp->err != nfs_ok)
1978
				break;
1979 1980 1981 1982 1983 1984

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1985 1986 1987 1988
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1989
		offset = vfs_llseek(file, 0, SEEK_CUR);
1990 1991 1992
	}

	free_page((unsigned long)(buf.dirent));
1993 1994 1995

	if (host_err)
		return nfserrno(host_err);
1996 1997 1998

	*offsetp = offset;
	return cdp->err;
1999 2000
}

L
Linus Torvalds 已提交
2001 2002 2003 2004
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
2005
__be32
L
Linus Torvalds 已提交
2006
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
2007
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
2008
{
2009
	__be32		err;
L
Linus Torvalds 已提交
2010 2011 2012
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
2013
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
2014 2015 2016 2017 2018 2019 2020 2021 2022
	if (err)
		goto out;

	offset = vfs_llseek(file, offset, 0);
	if (offset < 0) {
		err = nfserrno((int)offset);
		goto out_close;
	}

2023
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036

	if (err == nfserr_eof || err == nfserr_toosmall)
		err = nfs_ok; /* can still be found in ->err */
out_close:
	nfsd_close(file);
out:
	return err;
}

/*
 * Get file system stats
 * N.B. After this call fhp needs an fh_put
 */
2037
__be32
2038
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
2039
{
2040 2041 2042
	__be32 err;

	err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
2043 2044 2045 2046 2047 2048 2049 2050
	if (!err) {
		struct path path = {
			.mnt	= fhp->fh_export->ex_path.mnt,
			.dentry	= fhp->fh_dentry,
		};
		if (vfs_statfs(&path, stat))
			err = nfserr_io;
	}
L
Linus Torvalds 已提交
2051 2052 2053
	return err;
}

J
J. Bruce Fields 已提交
2054
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
2055
{
J
J. Bruce Fields 已提交
2056
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
2057 2058
}

L
Linus Torvalds 已提交
2059 2060 2061
/*
 * Check for a user's access permissions to this inode.
 */
2062
__be32
2063 2064
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
2065 2066 2067 2068
{
	struct inode	*inode = dentry->d_inode;
	int		err;

2069
	if ((acc & NFSD_MAY_MASK) == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
2070 2071 2072 2073
		return 0;
#if 0
	dprintk("nfsd: permission 0x%x%s%s%s%s%s%s%s mode 0%o%s%s%s\n",
		acc,
M
Miklos Szeredi 已提交
2074 2075 2076 2077 2078 2079 2080
		(acc & NFSD_MAY_READ)?	" read"  : "",
		(acc & NFSD_MAY_WRITE)?	" write" : "",
		(acc & NFSD_MAY_EXEC)?	" exec"  : "",
		(acc & NFSD_MAY_SATTR)?	" sattr" : "",
		(acc & NFSD_MAY_TRUNC)?	" trunc" : "",
		(acc & NFSD_MAY_LOCK)?	" lock"  : "",
		(acc & NFSD_MAY_OWNER_OVERRIDE)? " owneroverride" : "",
L
Linus Torvalds 已提交
2081 2082 2083
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2084
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2085
	dprintk("      owner %d/%d user %d/%d\n",
2086
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2087 2088 2089 2090 2091 2092
#endif

	/* Normally we reject any write/sattr etc access on a read-only file
	 * system.  But if it is IRIX doing check on write-access for a 
	 * device special file, we ignore rofs.
	 */
M
Miklos Szeredi 已提交
2093 2094
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2095 2096
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2097
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2098
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2099 2100
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2101
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2102 2103
		return nfserr_perm;

M
Miklos Szeredi 已提交
2104
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2105 2106 2107 2108 2109 2110 2111
		/* If we cannot rely on authentication in NLM requests,
		 * just allow locks, otherwise require read permission, or
		 * ownership
		 */
		if (exp->ex_flags & NFSEXP_NOAUTHNLM)
			return 0;
		else
M
Miklos Szeredi 已提交
2112
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127
	}
	/*
	 * The file owner always gets access permission for accesses that
	 * would normally be checked at open time. This is to make
	 * file access work even when the client has done a fchmod(fd, 0).
	 *
	 * However, `cp foo bar' should fail nevertheless when bar is
	 * readonly. A sensible way to do this might be to reject all
	 * attempts to truncate a read-only file, because a creat() call
	 * always implies file truncation.
	 * ... but this isn't really fair.  A process may reasonably call
	 * ftruncate on an open file descriptor on a file with perm 000.
	 * We must trust the client to do permission checking - using "ACCESS"
	 * with NFSv3.
	 */
M
Miklos Szeredi 已提交
2128
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2129
	    inode->i_uid == current_fsuid())
L
Linus Torvalds 已提交
2130 2131
		return 0;

M
Miklos Szeredi 已提交
2132
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2133
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2134 2135 2136

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
2137 2138
	     (acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE) ||
	      acc == (NFSD_MAY_READ | NFSD_MAY_READ_IF_EXEC)))
2139
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
2140 2141 2142 2143 2144 2145 2146

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2147 2148 2149
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2150
	dprintk("nfsd: freeing readahead buffers.\n");
2151 2152 2153 2154 2155 2156 2157 2158 2159 2160

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
		raparm = raparm_hash[i].pb_head;
		while(raparm) {
			last_raparm = raparm;
			raparm = raparm->p_next;
			kfree(last_raparm);
		}
		raparm_hash[i].pb_head = NULL;
	}
L
Linus Torvalds 已提交
2161 2162 2163 2164 2165 2166 2167 2168
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2169 2170
	int	j = 0;
	int	nperbucket;
2171
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2172

2173

2174
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2175
		return 0;
2176 2177 2178 2179
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2180 2181

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2182 2183

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2184
		spin_lock_init(&raparm_hash[i].pb_lock);
2185 2186 2187 2188 2189 2190 2191 2192 2193

		raparm = &raparm_hash[i].pb_head;
		for (j = 0; j < nperbucket; j++) {
			*raparm = kzalloc(sizeof(struct raparms), GFP_KERNEL);
			if (!*raparm)
				goto out_nomem;
			raparm = &(*raparm)->p_next;
		}
		*raparm = NULL;
2194 2195
	}

L
Linus Torvalds 已提交
2196 2197
	nfsdstats.ra_size = cache_size;
	return 0;
2198 2199 2200 2201 2202

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2203
}
2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214

#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
struct posix_acl *
nfsd_get_posix_acl(struct svc_fh *fhp, int type)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	char *name;
	void *value = NULL;
	ssize_t size;
	struct posix_acl *acl;

2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
	if (!IS_POSIXACL(inode))
		return ERR_PTR(-EOPNOTSUPP);

	switch (type) {
	case ACL_TYPE_ACCESS:
		name = POSIX_ACL_XATTR_ACCESS;
		break;
	case ACL_TYPE_DEFAULT:
		name = POSIX_ACL_XATTR_DEFAULT;
		break;
	default:
2226 2227 2228
		return ERR_PTR(-EOPNOTSUPP);
	}

2229 2230 2231
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246

	acl = posix_acl_from_xattr(value, size);
	kfree(value);
	return acl;
}

int
nfsd_set_posix_acl(struct svc_fh *fhp, int type, struct posix_acl *acl)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	char *name;
	void *value = NULL;
	size_t size;
	int error;

A
Al Viro 已提交
2247
	if (!IS_POSIXACL(inode) ||
2248 2249 2250 2251
	    !inode->i_op->setxattr || !inode->i_op->removexattr)
		return -EOPNOTSUPP;
	switch(type) {
		case ACL_TYPE_ACCESS:
2252
			name = POSIX_ACL_XATTR_ACCESS;
2253 2254
			break;
		case ACL_TYPE_DEFAULT:
2255
			name = POSIX_ACL_XATTR_DEFAULT;
2256 2257 2258 2259 2260 2261
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2262
		size = posix_acl_xattr_size(acl->a_count);
2263 2264 2265
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2266 2267
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2268
			goto getout;
2269
		size = error;
2270 2271 2272
	} else
		size = 0;

A
Al Viro 已提交
2273
	error = fh_want_write(fhp);
2274 2275
	if (error)
		goto getout;
2276
	if (size)
2277
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2278 2279 2280 2281
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2282
			error = vfs_removexattr(fhp->fh_dentry, name);
2283 2284 2285 2286
			if (error == -ENODATA)
				error = 0;
		}
	}
A
Al Viro 已提交
2287
	fh_drop_write(fhp);
2288 2289 2290 2291 2292 2293

getout:
	kfree(value);
	return error;
}
#endif  /* defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) */