vfs.c 53.4 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
#define MSNFS	/* HACK HACK */
/*
 * 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>
19
#include <linux/splice.h>
L
Linus Torvalds 已提交
20 21 22
#include <linux/fcntl.h>
#include <linux/namei.h>
#include <linux/delay.h>
R
Robert Love 已提交
23
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
24 25
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
26 27
#include <linux/jhash.h>
#include <linux/ima.h>
28
#include <linux/slab.h>
29
#include <asm/uaccess.h>
30 31
#include <linux/exportfs.h>
#include <linux/writeback.h>
32 33 34 35 36

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

37
#ifdef CONFIG_NFSD_V4
L
Linus Torvalds 已提交
38 39 40 41
#include <linux/nfs4_acl.h>
#include <linux/nfsd_idmap.h>
#endif /* CONFIG_NFSD_V4 */

42 43
#include "nfsd.h"
#include "vfs.h"
L
Linus Torvalds 已提交
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61

#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;
62
	unsigned int		p_hindex;
L
Linus Torvalds 已提交
63 64
};

65 66 67 68 69 70 71 72 73
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
74 75 76 77

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
78
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
L
Linus Torvalds 已提交
79 80 81 82 83 84 85 86
 *  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
87 88
	struct path path = {.mnt = mntget(exp->ex_path.mnt),
			    .dentry = dget(dentry)};
89
	int err = 0;
L
Linus Torvalds 已提交
90

A
Al Viro 已提交
91
	while (d_mountpoint(path.dentry) && follow_down(&path))
A
Al Viro 已提交
92
		;
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)
168 169 170 171 172 173 174 175
{
	if (d_mountpoint(dentry))
		return 1;
	if (!(exp->ex_flags & NFSEXP_V4ROOT))
		return 0;
	return dentry->d_inode != NULL;
}

176
__be32
177
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
178
		   const char *name, unsigned int len,
179
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
180 181 182 183
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
184 185
	__be32			err;
	int			host_err;
L
Linus Torvalds 已提交
186 187 188 189

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

	/* Obtain dentry and export. */
M
Miklos Szeredi 已提交
190
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
191 192 193 194 195 196 197 198 199 200 201
	if (err)
		return err;

	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);
202
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
203
			dentry = dget_parent(dparent);
204
		else if (!EX_NOHIDE(exp) && !nfsd_v4client(rqstp))
L
Linus Torvalds 已提交
205 206 207
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
208 209
			host_err = nfsd_lookup_parent(rqstp, dparent, &exp, &dentry);
			if (host_err)
L
Linus Torvalds 已提交
210 211 212 213 214
				goto out_nfserr;
		}
	} else {
		fh_lock(fhp);
		dentry = lookup_one_len(name, dparent, len);
215
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
216 217 218 219 220
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
221
		if (nfsd_mountpoint(dentry, exp)) {
222
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
223 224 225 226 227
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
	*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,
251
				unsigned int len, struct svc_fh *resfh)
252 253 254 255 256 257 258 259
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
260 261 262
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
263 264 265 266 267 268 269
	/*
	 * 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;
270
out:
L
Linus Torvalds 已提交
271 272 273 274 275
	dput(dentry);
	exp_put(exp);
	return err;
}

276 277 278 279 280 281 282 283 284 285 286 287
/*
 * 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
288 289 290
	if (export_ops->commit_metadata)
		return export_ops->commit_metadata(inode);
	return sync_inode_metadata(inode, 1);
291
}
292

L
Linus Torvalds 已提交
293 294 295 296
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
297
__be32
L
Linus Torvalds 已提交
298 299 300 301 302
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
303
	int		accmode = NFSD_MAY_SATTR;
L
Linus Torvalds 已提交
304
	int		ftype = 0;
305 306
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
307 308 309
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
310
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
311 312 313 314 315
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
316
	if (err)
L
Linus Torvalds 已提交
317 318 319 320 321
		goto out;

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

322 323 324 325 326 327 328
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

329 330
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
331 332 333 334 335 336 337 338 339 340 341 342
	 * 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)
343 344 345 346 347 348 349 350
	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
351
		 */
352 353 354 355 356 357 358 359 360 361 362 363
		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
364 365
	}
	    
366 367 368 369
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
370 371
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
372 373
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
374 375 376 377 378 379 380 381
			if (err)
				goto out;
		}

		/*
		 * If we are changing the size of the file, then
		 * we need to break all leases.
		 */
382
		host_err = break_lease(inode, O_WRONLY | O_NONBLOCK);
383
		if (host_err) /* ENOMEM or EWOULDBLOCK */
L
Linus Torvalds 已提交
384 385
			goto out_nfserr;

386 387
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
388 389 390
			goto out_nfserr;

		size_change = 1;
391 392
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
393 394 395 396 397
			put_write_access(inode);
			goto out_nfserr;
		}
	}

398
	/* sanitize the mode change */
L
Linus Torvalds 已提交
399 400
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
401
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
402 403 404
	}

	/* Revoke setuid/setgid on chown */
405 406 407
	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))) {
408 409 410
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
411
			iap->ia_mode &= ~S_ISUID;
412 413 414 415 416
			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);
417
		}
L
Linus Torvalds 已提交
418 419 420 421 422 423 424 425 426
	}

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
		fh_lock(fhp);
427 428
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
429 430 431 432 433
		fh_unlock(fhp);
	}
	if (size_change)
		put_write_access(inode);
	if (!err)
434
		commit_metadata(fhp);
L
Linus Torvalds 已提交
435 436 437 438
out:
	return err;

out_nfserr:
439
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
440 441 442
	goto out;
}

443 444 445 446 447 448
#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;
449
	ssize_t ret;
450 451 452 453

	buflen = vfs_getxattr(dentry, key, NULL, 0);
	if (buflen <= 0)
		return buflen;
L
Linus Torvalds 已提交
454

455 456 457 458
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

459 460 461 462
	ret = vfs_getxattr(dentry, key, *buf, buflen);
	if (ret < 0)
		kfree(*buf);
	return ret;
463 464 465 466
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
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;
	}

487
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
488 489 490 491 492
out:
	kfree(buf);
	return error;
}

493
__be32
L
Linus Torvalds 已提交
494 495 496
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
497 498
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
499 500 501 502 503 504
	struct dentry *dentry;
	struct inode *inode;
	struct posix_acl *pacl = NULL, *dpacl = NULL;
	unsigned int flags = 0;

	/* Get inode */
M
Miklos Szeredi 已提交
505
	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
506
	if (error)
507
		return error;
L
Linus Torvalds 已提交
508 509 510 511 512 513

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

514 515
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
516
		return nfserr_attrnotsupp;
517
	} else if (host_error < 0)
L
Linus Torvalds 已提交
518 519
		goto out_nfserr;

520 521
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
522
		goto out_release;
L
Linus Torvalds 已提交
523

524
	if (S_ISDIR(inode->i_mode))
525
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
526

527
out_release:
L
Linus Torvalds 已提交
528 529 530
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
531
	if (host_error == -EOPNOTSUPP)
532
		return nfserr_attrnotsupp;
533
	else
534
		return nfserrno(host_error);
L
Linus Torvalds 已提交
535 536 537 538 539
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
540
	void *buf = NULL;
L
Linus Torvalds 已提交
541
	struct posix_acl *pacl = NULL;
542
	int buflen;
L
Linus Torvalds 已提交
543

544 545 546 547 548
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
549 550 551 552 553 554 555 556 557 558 559 560 561 562

	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;

563
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
564 565 566 567 568 569 570 571 572
	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)) {
573
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594
		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;
}

J
J. Bruce Fields 已提交
595
#endif /* defined(CONFIG_NFSD_V4) */
L
Linus Torvalds 已提交
596 597 598 599 600 601 602 603 604 605

#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
606 607 608 609
    {	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
610 611 612 613 614

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
615 616 617 618 619
    {	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
620 621 622 623 624 625 626 627 628 629 630 631

    {	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
632 633 634 635
    {	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
636 637 638 639

    {	0,			0				}
};

640
__be32
L
Linus Torvalds 已提交
641 642 643 644 645 646
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;
647
	__be32			error;
L
Linus Torvalds 已提交
648

M
Miklos Szeredi 已提交
649
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
	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) {
667
			__be32 err2;
L
Linus Torvalds 已提交
668 669 670

			sresult |= map->access;

671
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705
			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 */



/*
 * 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
 */
706
__be32
L
Linus Torvalds 已提交
707 708 709 710 711
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
			int access, struct file **filp)
{
	struct dentry	*dentry;
	struct inode	*inode;
712 713
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
714
	int		host_err = 0;
L
Linus Torvalds 已提交
715

716 717
	validate_process_creds();

L
Linus Torvalds 已提交
718 719 720 721 722
	/*
	 * 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
723
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
724 725 726 727 728 729 730 731 732 733
	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
734
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
735
		goto out;
J
J. Bruce Fields 已提交
736 737 738 739 740 741
	/*
	 * 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
742 743 744 745 746 747 748 749 750
		goto out;

	if (!inode->i_fop)
		goto out;

	/*
	 * Check to see if there are any leases on this file.
	 * This may block while leases are broken.
	 */
751 752
	if (!(access & NFSD_MAY_NOT_BREAK_LEASE))
		host_err = break_lease(inode, O_NONBLOCK | ((access & NFSD_MAY_WRITE) ? O_WRONLY : 0));
753
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
754 755
		goto out_nfserr;

M
Miklos Szeredi 已提交
756 757
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
758 759 760
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
761
	}
762
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
763
			    flags, current_cred());
L
Linus Torvalds 已提交
764
	if (IS_ERR(*filp))
765
		host_err = PTR_ERR(*filp);
766 767
	else
		host_err = ima_file_check(*filp, access);
L
Linus Torvalds 已提交
768
out_nfserr:
769
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
770
out:
771
	validate_process_creds();
L
Linus Torvalds 已提交
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
	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;
794 795 796 797 798
	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
799

800 801
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
802 803 804 805 806 807 808 809
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
	depth = nfsdstats.ra_size*11/10;
	if (!frap) {	
810
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
811 812 813 814 815 816 817
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
818
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
819
found:
820
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
821
		*rap = ra->p_next;
822 823
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
824 825 826
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
827
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
828 829 830 831
	return ra;
}

/*
832 833 834
 * 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
835 836
 */
static int
837 838
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
839
{
840
	struct svc_rqst *rqstp = sd->u.data;
841
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
842 843 844
	struct page *page = buf->page;
	size_t size;
	int ret;
L
Linus Torvalds 已提交
845

846
	ret = buf->ops->confirm(pipe, buf);
847 848 849 850
	if (unlikely(ret))
		return ret;

	size = sd->len;
L
Linus Torvalds 已提交
851 852 853

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
854 855 856
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
857
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
858
		rqstp->rq_res.page_len = size;
859
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
860
		get_page(page);
861 862
		if (*pp)
			put_page(*pp);
863 864
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
865
		rqstp->rq_res.page_len += size;
866
	} else
L
Linus Torvalds 已提交
867 868 869 870 871
		rqstp->rq_res.page_len += size;

	return size;
}

872 873 874 875 876 877
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

878 879 880 881 882 883 884 885 886
static inline int svc_msnfs(struct svc_fh *ffhp)
{
#ifdef MSNFS
	return (ffhp->fh_export->ex_flags & NFSEXP_MSNFS);
#else
	return 0;
#endif
}

887
static __be32
L
Linus Torvalds 已提交
888 889 890 891 892
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)
{
	struct inode *inode;
	mm_segment_t	oldfs;
893 894
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
895 896

	err = nfserr_perm;
897
	inode = file->f_path.dentry->d_inode;
898 899

	if (svc_msnfs(fhp) && !lock_may_read(inode, offset, *count))
L
Linus Torvalds 已提交
900 901
		goto out;

902 903 904 905 906 907 908 909
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

910
		rqstp->rq_resused = 1;
911
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
912 913 914
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
915
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
916 917 918
		set_fs(oldfs);
	}

919 920 921
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
922
		err = 0;
923
		fsnotify_access(file);
L
Linus Torvalds 已提交
924
	} else 
925
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
926 927 928 929
out:
	return err;
}

930 931 932
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
933
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
934

935
	mutex_lock(&dentry->d_inode->i_mutex);
936
	notify_change(dentry, &ia);
937
	mutex_unlock(&dentry->d_inode->i_mutex);
938 939
}

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
/*
 * 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));
970
		err = vfs_fsync(file, 0);
971 972 973 974 975 976
	}
	last_ino = inode->i_ino;
	last_dev = inode->i_sb->s_dev;
	return err;
}

977
static __be32
L
Linus Torvalds 已提交
978 979
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
980
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
981 982 983 984 985
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
986 987
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
988
	int			stable = *stablep;
989
	int			use_wgather;
L
Linus Torvalds 已提交
990

991
#ifdef MSNFS
L
Linus Torvalds 已提交
992 993 994
	err = nfserr_perm;

	if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
995
		(!lock_may_write(file->f_path.dentry->d_inode, offset, *cnt)))
L
Linus Torvalds 已提交
996 997 998
		goto out;
#endif

999
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
1000 1001 1002 1003 1004 1005 1006 1007
	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().
1008
	 * When NFSv2 gathered writes have been configured for this volume,
L
Linus Torvalds 已提交
1009 1010
	 * flushing the data to disk is handled separately below.
	 */
1011
	use_wgather = (rqstp->rq_vers == 2) && EX_WGATHER(exp);
L
Linus Torvalds 已提交
1012

1013
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
1014 1015 1016 1017 1018 1019
	       stable = 2;
	       *stablep = 2; /* FILE_SYNC */
	}

	if (!EX_ISSYNC(exp))
		stable = 0;
1020
	if (stable && !use_wgather) {
J
Jonathan Corbet 已提交
1021
		spin_lock(&file->f_lock);
L
Linus Torvalds 已提交
1022
		file->f_flags |= O_SYNC;
J
Jonathan Corbet 已提交
1023 1024
		spin_unlock(&file->f_lock);
	}
L
Linus Torvalds 已提交
1025 1026 1027

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
1028
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1029
	set_fs(oldfs);
1030 1031 1032 1033
	if (host_err < 0)
		goto out_nfserr;
	*cnt = host_err;
	nfsdstats.io_write += host_err;
1034
	fsnotify_modify(file);
L
Linus Torvalds 已提交
1035 1036

	/* clear setuid/setgid flag after write */
1037
	if (inode->i_mode & (S_ISUID | S_ISGID))
1038
		kill_suid(dentry);
L
Linus Torvalds 已提交
1039

1040
	if (stable && use_wgather)
1041
		host_err = wait_for_concurrent_writes(file);
L
Linus Torvalds 已提交
1042

1043
out_nfserr:
1044
	dprintk("nfsd: write complete host_err=%d\n", host_err);
1045
	if (host_err >= 0)
L
Linus Torvalds 已提交
1046
		err = 0;
1047
	else
1048
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053 1054 1055 1056 1057
out:
	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
 */
1058
__be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
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 1086 1087 1088 1089 1090 1091 1092 1093
	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;
}

1094
/* As above, but use the provided file descriptor. */
1095
__be32
1096
nfsd_read_file(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
L
Linus Torvalds 已提交
1097 1098 1099
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1100
	__be32		err;
L
Linus Torvalds 已提交
1101 1102

	if (file) {
1103
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1104
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1105 1106 1107
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
1108 1109
	} 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
1110 1111 1112 1113 1114 1115 1116 1117 1118
out:
	return err;
}

/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1119
__be32
L
Linus Torvalds 已提交
1120
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1121
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1122 1123
		int *stablep)
{
1124
	__be32			err = 0;
L
Linus Torvalds 已提交
1125 1126

	if (file) {
1127
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1128
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1129 1130 1131 1132 1133
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1134
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
		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.
1150 1151 1152
 *
 * 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
1153 1154 1155 1156
 *
 * 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.
 */
1157
__be32
L
Linus Torvalds 已提交
1158 1159 1160 1161
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1162 1163
	loff_t		end = LLONG_MAX;
	__be32		err = nfserr_inval;
L
Linus Torvalds 已提交
1164

1165 1166 1167 1168 1169 1170 1171
	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
1172

1173 1174
	err = nfsd_open(rqstp, fhp, S_IFREG,
			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &file);
M
Miklos Szeredi 已提交
1175
	if (err)
1176
		goto out;
L
Linus Torvalds 已提交
1177
	if (EX_ISSYNC(fhp->fh_export)) {
1178
		int err2 = vfs_fsync_range(file, offset, end, 0);
1179 1180 1181 1182

		if (err2 != -EINVAL)
			err = nfserrno(err2);
		else
L
Linus Torvalds 已提交
1183 1184 1185 1186
			err = nfserr_notsupp;
	}

	nfsd_close(file);
1187
out:
L
Linus Torvalds 已提交
1188 1189 1190 1191
	return err;
}
#endif /* CONFIG_NFSD_V3 */

A
Adrian Bunk 已提交
1192
static __be32
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
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:
	 */
1205
	if (current_fsuid() != 0)
1206 1207 1208 1209 1210 1211
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226
/* 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
1227 1228 1229 1230 1231 1232 1233 1234
/*
 * 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
 */
1235
__be32
L
Linus Torvalds 已提交
1236 1237 1238 1239 1240 1241
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;
1242
	__be32		err;
1243
	__be32		err2;
1244
	int		host_err;
L
Linus Torvalds 已提交
1245 1246 1247 1248 1249 1250 1251 1252

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

M
Miklos Szeredi 已提交
1253
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260
	if (err)
		goto out;

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

	err = nfserr_notdir;
A
Al Viro 已提交
1261
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1262 1263 1264 1265 1266 1267 1268
		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
1269
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1270
		dchild = lookup_one_len(fname, dentry, flen);
1271
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
		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
1286
			err = nfserr_io;
L
Linus Torvalds 已提交
1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
			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;

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
	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;
	}

	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1315 1316 1317
	/*
	 * Get the dir op function pointer.
	 */
1318
	err = 0;
L
Linus Torvalds 已提交
1319 1320
	switch (type) {
	case S_IFREG:
1321
		host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1322 1323
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
L
Linus Torvalds 已提交
1324 1325
		break;
	case S_IFDIR:
1326
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1327 1328 1329 1330 1331
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1332
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1333 1334
		break;
	}
1335 1336
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1337
		goto out_nfserr;
1338
	}
L
Linus Torvalds 已提交
1339

1340
	err = nfsd_create_setattr(rqstp, resfhp, iap);
L
Linus Torvalds 已提交
1341

1342 1343 1344 1345 1346 1347
	/*
	 * 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));
1348 1349
	if (err2)
		err = err2;
1350
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
	/*
	 * 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:
1362
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1363 1364 1365 1366 1367 1368 1369
	goto out;
}

#ifdef CONFIG_NFSD_V3
/*
 * NFSv3 version of nfsd_create
 */
1370
__be32
L
Linus Torvalds 已提交
1371 1372 1373
nfsd_create_v3(struct svc_rqst *rqstp, struct svc_fh *fhp,
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1374
	        int *truncp, int *created)
L
Linus Torvalds 已提交
1375 1376 1377
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1378 1379
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
	__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;
M
Miklos Szeredi 已提交
1390
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1391 1392 1393 1394 1395 1396 1397 1398 1399
	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
1400
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1401
		goto out;
P
Peter Zijlstra 已提交
1402
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1403 1404 1405 1406 1407

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1408
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1409 1410 1411 1412 1413 1414 1415 1416
	if (IS_ERR(dchild))
		goto out_nfserr;

	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
	if (err)
		goto out;

	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1417
		/* solaris7 gets confused (bugid 4218508) if these have
1418 1419 1420 1421
		 * 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
1422 1423 1424 1425 1426
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1427 1428 1429
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1430 1431 1432 1433 1434 1435 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
	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;
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
1461
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1462 1463 1464
		goto out;
	}

1465
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1466 1467
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1468
		goto out_nfserr;
1469
	}
1470 1471
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1472

1473 1474
	nfsd_check_ignore_resizing(iap);

L
Linus Torvalds 已提交
1475
	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1476
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1477
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1478
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1479 1480 1481 1482 1483 1484 1485 1486
		/* 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:
1487 1488 1489 1490 1491 1492 1493
	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));
1494

1495
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
1496 1497 1498 1499 1500
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1509
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514 1515 1516 1517 1518
	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
 */
1519
__be32
L
Linus Torvalds 已提交
1520 1521 1522 1523 1524
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1525 1526
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1527

M
Miklos Szeredi 已提交
1528
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534 1535
	if (err)
		goto out;

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

	err = nfserr_inval;
A
Al Viro 已提交
1536
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1537 1538
		goto out;

1539
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1540 1541 1542 1543 1544
	/* 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);
1545
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1546 1547
	set_fs(oldfs);

1548
	if (host_err < 0)
L
Linus Torvalds 已提交
1549
		goto out_nfserr;
1550
	*lenp = host_err;
L
Linus Torvalds 已提交
1551 1552 1553 1554 1555
	err = 0;
out:
	return err;

out_nfserr:
1556
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1557 1558 1559 1560 1561 1562 1563
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1564
__be32
L
Linus Torvalds 已提交
1565 1566 1567 1568 1569 1570 1571
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;
1572 1573
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1574 1575 1576 1577 1578 1579 1580 1581

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

M
Miklos Szeredi 已提交
1582
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1583 1584 1585 1586 1587
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1588
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1589 1590 1591
	if (IS_ERR(dnew))
		goto out_nfserr;

1592 1593 1594 1595
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1596 1597 1598
	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1599
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1600 1601 1602
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1603
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1604 1605 1606
			kfree(path_alloced);
		}
	} else
1607
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
1608
	err = nfserrno(host_err);
1609 1610
	if (!err)
		err = nfserrno(commit_metadata(fhp));
L
Linus Torvalds 已提交
1611 1612
	fh_unlock(fhp);

1613 1614
	mnt_drop_write(fhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1615 1616 1617 1618 1619 1620 1621
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1622
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1623 1624 1625 1626 1627 1628 1629
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1630
__be32
L
Linus Torvalds 已提交
1631 1632 1633 1634
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
1635
	struct inode	*dirp;
1636 1637
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1638

M
Miklos Szeredi 已提交
1639
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1640 1641
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1642
	err = fh_verify(rqstp, tfhp, -S_IFDIR, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652
	if (err)
		goto out;

	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

P
Peter Zijlstra 已提交
1653
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1654 1655 1656 1657
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1658
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1659 1660 1661 1662 1663
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;

1664 1665 1666 1667 1668
	host_err = mnt_want_write(tfhp->fh_export->ex_path.mnt);
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1669 1670
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
1671 1672 1673
		err = nfserrno(commit_metadata(ffhp));
		if (!err)
			err = nfserrno(commit_metadata(tfhp));
L
Linus Torvalds 已提交
1674
	} else {
1675
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1676 1677
			err = nfserr_acces;
		else
1678
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1679
	}
1680 1681
	mnt_drop_write(tfhp->fh_export->ex_path.mnt);
out_dput:
L
Linus Torvalds 已提交
1682
	dput(dnew);
1683 1684
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1685 1686 1687 1688
out:
	return err;

out_nfserr:
1689
	err = nfserrno(host_err);
1690
	goto out_unlock;
L
Linus Torvalds 已提交
1691 1692 1693 1694 1695 1696
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1697
__be32
L
Linus Torvalds 已提交
1698 1699 1700 1701 1702
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;
1703 1704
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1705

M
Miklos Szeredi 已提交
1706
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1707 1708
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1709
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
	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;
1720
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
		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);
1735
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1736 1737 1738
	if (IS_ERR(odentry))
		goto out_nfserr;

1739
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1740 1741
	if (!odentry->d_inode)
		goto out_dput_old;
1742
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1743 1744 1745 1746
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1747
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1748 1749
	if (IS_ERR(ndentry))
		goto out_dput_old;
1750
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1751 1752 1753
	if (ndentry == trap)
		goto out_dput_new;

1754
	if (svc_msnfs(ffhp) &&
L
Linus Torvalds 已提交
1755 1756
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
1757
			host_err = -EPERM;
1758 1759 1760 1761 1762 1763 1764 1765 1766 1767
			goto out_dput_new;
	}

	host_err = -EXDEV;
	if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
		goto out_dput_new;
	host_err = mnt_want_write(ffhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_dput_new;

1768
	host_err = vfs_rename(fdir, odentry, tdir, ndentry);
1769 1770
	if (!host_err) {
		host_err = commit_metadata(tfhp);
1771
		if (!host_err)
1772
			host_err = commit_metadata(ffhp);
L
Linus Torvalds 已提交
1773 1774
	}

1775 1776
	mnt_drop_write(ffhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1777 1778 1779 1780 1781
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1782
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800

	/* 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
 */
1801
__be32
L
Linus Torvalds 已提交
1802 1803 1804 1805 1806
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1807 1808
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1809 1810 1811 1812

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1813
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1814 1815 1816
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1817
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1818 1819 1820 1821
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1822
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834
	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;

1835 1836 1837 1838
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1839 1840 1841 1842
	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
1843
			host_err = -EPERM;
L
Linus Torvalds 已提交
1844 1845
		} else
#endif
1846
		host_err = vfs_unlink(dirp, rdentry);
L
Linus Torvalds 已提交
1847
	} else { /* It's RMDIR */
1848
		host_err = vfs_rmdir(dirp, rdentry);
L
Linus Torvalds 已提交
1849 1850 1851 1852
	}

	dput(rdentry);

1853 1854
	if (!host_err)
		host_err = commit_metadata(fhp);
L
Linus Torvalds 已提交
1855

1856
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1857
out_nfserr:
1858
	err = nfserrno(host_err);
1859 1860
out:
	return err;
L
Linus Torvalds 已提交
1861 1862
}

1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880
/*
 * 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;
1881
	int		full;
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891
};

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));
1892 1893
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1894
		return -EINVAL;
1895
	}
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906

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

	return 0;
}

1907 1908
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1909
{
1910 1911
	struct readdir_data buf;
	struct buffered_dirent *de;
1912
	int host_err;
1913 1914
	int size;
	loff_t offset;
1915

1916 1917
	buf.dirent = (void *)__get_free_page(GFP_KERNEL);
	if (!buf.dirent)
1918
		return nfserrno(-ENOMEM);
1919 1920

	offset = *offsetp;
1921

1922
	while (1) {
1923
		struct inode *dir_inode = file->f_path.dentry->d_inode;
1924 1925
		unsigned int reclen;

1926
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1927
		buf.used = 0;
1928
		buf.full = 0;
1929 1930

		host_err = vfs_readdir(file, nfsd_buffered_filldir, &buf);
1931 1932 1933 1934
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1935 1936 1937 1938 1939 1940 1941
			break;

		size = buf.used;

		if (!size)
			break;

1942 1943 1944 1945 1946 1947 1948 1949 1950
		/*
		 * 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;

1951 1952 1953 1954 1955 1956
		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))
1957
				break;
1958 1959

			if (cdp->err != nfs_ok)
1960
				break;
1961 1962 1963 1964 1965 1966

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1967 1968 1969 1970
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1971
		offset = vfs_llseek(file, 0, SEEK_CUR);
1972 1973 1974
	}

	free_page((unsigned long)(buf.dirent));
1975 1976 1977

	if (host_err)
		return nfserrno(host_err);
1978 1979 1980

	*offsetp = offset;
	return cdp->err;
1981 1982
}

L
Linus Torvalds 已提交
1983 1984 1985 1986
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1987
__be32
L
Linus Torvalds 已提交
1988
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1989
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1990
{
1991
	__be32		err;
L
Linus Torvalds 已提交
1992 1993 1994
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
1995
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1996 1997 1998 1999 2000 2001 2002 2003 2004
	if (err)
		goto out;

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

2005
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

	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
 */
2019
__be32
2020
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
2021
{
2022 2023 2024
	__be32 err;

	err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
2025 2026 2027 2028 2029 2030 2031 2032
	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
2033 2034 2035
	return err;
}

J
J. Bruce Fields 已提交
2036
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
2037
{
J
J. Bruce Fields 已提交
2038
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
2039 2040
}

L
Linus Torvalds 已提交
2041 2042 2043
/*
 * Check for a user's access permissions to this inode.
 */
2044
__be32
2045 2046
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
2047 2048 2049 2050
{
	struct inode	*inode = dentry->d_inode;
	int		err;

M
Miklos Szeredi 已提交
2051
	if (acc == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
2052 2053 2054 2055
		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
2056 2057 2058 2059 2060 2061 2062
		(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
2063 2064 2065
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2066
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2067
	dprintk("      owner %d/%d user %d/%d\n",
2068
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2069 2070 2071 2072 2073 2074
#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
2075 2076
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2077 2078
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2079
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2080
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2081 2082
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2083
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2084 2085
		return nfserr_perm;

M
Miklos Szeredi 已提交
2086
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2087 2088 2089 2090 2091 2092 2093
		/* 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
2094
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109
	}
	/*
	 * 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
2110
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2111
	    inode->i_uid == current_fsuid())
L
Linus Torvalds 已提交
2112 2113
		return 0;

M
Miklos Szeredi 已提交
2114
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2115
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2116 2117 2118

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
M
Miklos Szeredi 已提交
2119
	    acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE))
2120
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
2121 2122 2123 2124 2125 2126 2127

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2128 2129 2130
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2131
	dprintk("nfsd: freeing readahead buffers.\n");
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141

	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
2142 2143 2144 2145 2146 2147 2148 2149
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2150 2151
	int	j = 0;
	int	nperbucket;
2152
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2153

2154

2155
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2156
		return 0;
2157 2158 2159 2160
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2161 2162

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2163 2164

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2165
		spin_lock_init(&raparm_hash[i].pb_lock);
2166 2167 2168 2169 2170 2171 2172 2173 2174

		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;
2175 2176
	}

L
Linus Torvalds 已提交
2177 2178
	nfsdstats.ra_size = cache_size;
	return 0;
2179 2180 2181 2182 2183

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2184
}
2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195

#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;

2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206
	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:
2207 2208 2209
		return ERR_PTR(-EOPNOTSUPP);
	}

2210 2211 2212
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227

	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
2228
	if (!IS_POSIXACL(inode) ||
2229 2230 2231 2232
	    !inode->i_op->setxattr || !inode->i_op->removexattr)
		return -EOPNOTSUPP;
	switch(type) {
		case ACL_TYPE_ACCESS:
2233
			name = POSIX_ACL_XATTR_ACCESS;
2234 2235
			break;
		case ACL_TYPE_DEFAULT:
2236
			name = POSIX_ACL_XATTR_DEFAULT;
2237 2238 2239 2240 2241 2242
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2243
		size = posix_acl_xattr_size(acl->a_count);
2244 2245 2246
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2247 2248
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2249
			goto getout;
2250
		size = error;
2251 2252 2253
	} else
		size = 0;

2254 2255 2256
	error = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (error)
		goto getout;
2257
	if (size)
2258
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2259 2260 2261 2262
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2263
			error = vfs_removexattr(fhp->fh_dentry, name);
2264 2265 2266 2267
			if (error == -ENODATA)
				error = 0;
		}
	}
2268
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
2269 2270 2271 2272 2273 2274

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