vfs.c 53.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
#define MSNFS	/* HACK HACK */
/*
 * linux/fs/nfsd/vfs.c
 *
 * 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/string.h>
#include <linux/time.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/mount.h>
#include <linux/major.h>
26
#include <linux/splice.h>
L
Linus Torvalds 已提交
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
#include <linux/proc_fs.h>
#include <linux/stat.h>
#include <linux/fcntl.h>
#include <linux/net.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/in.h>
#include <linux/module.h>
#include <linux/namei.h>
#include <linux/vfs.h>
#include <linux/delay.h>
#include <linux/sunrpc/svc.h>
#include <linux/nfsd/nfsd.h>
#ifdef CONFIG_NFSD_V3
#include <linux/nfs3.h>
#include <linux/nfsd/xdr3.h>
#endif /* CONFIG_NFSD_V3 */
#include <linux/nfsd/nfsfh.h>
#include <linux/quotaops.h>
R
Robert Love 已提交
47
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
48 49 50
#include <linux/posix_acl.h>
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
51
#ifdef CONFIG_NFSD_V4
L
Linus Torvalds 已提交
52 53 54 55 56
#include <linux/nfs4.h>
#include <linux/nfs4_acl.h>
#include <linux/nfsd_idmap.h>
#include <linux/security.h>
#endif /* CONFIG_NFSD_V4 */
57
#include <linux/jhash.h>
M
Mimi Zohar 已提交
58
#include <linux/ima.h>
L
Linus Torvalds 已提交
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78

#include <asm/uaccess.h>

#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;
79
	unsigned int		p_hindex;
L
Linus Torvalds 已提交
80 81
};

82 83 84 85 86 87 88 89 90
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
91 92 93 94

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
95
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
L
Linus Torvalds 已提交
96 97 98 99 100 101 102 103
 *  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
104 105
	struct path path = {.mnt = mntget(exp->ex_path.mnt),
			    .dentry = dget(dentry)};
106
	int err = 0;
L
Linus Torvalds 已提交
107

A
Al Viro 已提交
108 109 110
	while (follow_down(&path.mnt, &path.dentry) &&
	       d_mountpoint(path.dentry))
		;
L
Linus Torvalds 已提交
111

A
Al Viro 已提交
112
	exp2 = rqst_exp_get_by_name(rqstp, &path);
L
Linus Torvalds 已提交
113
	if (IS_ERR(exp2)) {
114 115
		if (PTR_ERR(exp2) != -ENOENT)
			err = PTR_ERR(exp2);
A
Al Viro 已提交
116
		path_put(&path);
L
Linus Torvalds 已提交
117 118
		goto out;
	}
119
	if ((exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
L
Linus Torvalds 已提交
120
		/* successfully crossed mount point */
A
Al Viro 已提交
121
		/*
A
Al Viro 已提交
122 123 124 125
		 * 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
126
		 */
A
Al Viro 已提交
127 128
		*dpp = path.dentry;
		path.dentry = dentry;
A
Al Viro 已提交
129
		*expp = exp2;
A
Al Viro 已提交
130
		exp2 = exp;
L
Linus Torvalds 已提交
131
	}
A
Al Viro 已提交
132 133
	path_put(&path);
	exp_put(exp2);
L
Linus Torvalds 已提交
134 135 136 137
out:
	return err;
}

138
__be32
139
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
140
		   const char *name, unsigned int len,
141
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
142 143 144 145
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
146 147
	__be32			err;
	int			host_err;
L
Linus Torvalds 已提交
148 149 150 151

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

	/* Obtain dentry and export. */
M
Miklos Szeredi 已提交
152
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163
	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);
164
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
165
			dentry = dget_parent(dparent);
166
		else if (!EX_NOHIDE(exp))
L
Linus Torvalds 已提交
167 168 169 170 171
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
			struct svc_export *exp2 = NULL;
			struct dentry *dp;
A
Al Viro 已提交
172 173 174 175
			struct path path = {.mnt = mntget(exp->ex_path.mnt),
					    .dentry = dget(dparent)};

			while (path.dentry == path.mnt->mnt_root &&
A
Al Viro 已提交
176
			       follow_up(&path))
L
Linus Torvalds 已提交
177
				;
A
Al Viro 已提交
178 179 180
			dp = dget_parent(path.dentry);
			dput(path.dentry);
			path.dentry = dp;
L
Linus Torvalds 已提交
181

A
Al Viro 已提交
182
			exp2 = rqst_exp_parent(rqstp, &path);
183 184 185
			if (PTR_ERR(exp2) == -ENOENT) {
				dentry = dget(dparent);
			} else if (IS_ERR(exp2)) {
186
				host_err = PTR_ERR(exp2);
A
Al Viro 已提交
187
				path_put(&path);
L
Linus Torvalds 已提交
188 189
				goto out_nfserr;
			} else {
A
Al Viro 已提交
190
				dentry = dget(path.dentry);
L
Linus Torvalds 已提交
191 192 193
				exp_put(exp);
				exp = exp2;
			}
A
Al Viro 已提交
194
			path_put(&path);
L
Linus Torvalds 已提交
195 196 197 198
		}
	} else {
		fh_lock(fhp);
		dentry = lookup_one_len(name, dparent, len);
199
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
200 201 202 203 204 205
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
		if (d_mountpoint(dentry)) {
206
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
207 208 209 210 211
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
	*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,
235
				unsigned int len, struct svc_fh *resfh)
236 237 238 239 240 241 242 243
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
244 245 246
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
247 248 249 250 251 252 253
	/*
	 * 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;
254
out:
L
Linus Torvalds 已提交
255 256 257 258 259
	dput(dentry);
	exp_put(exp);
	return err;
}

260

L
Linus Torvalds 已提交
261 262 263 264
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
265
__be32
L
Linus Torvalds 已提交
266 267 268 269 270
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
271
	int		accmode = NFSD_MAY_SATTR;
L
Linus Torvalds 已提交
272
	int		ftype = 0;
273 274
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
275 276 277
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
278
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
279 280 281 282 283
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
284
	if (err)
L
Linus Torvalds 已提交
285 286 287 288 289
		goto out;

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

290 291 292 293 294 295 296
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

297 298
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
299 300 301 302 303 304 305 306 307 308 309 310
	 * 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)
311 312 313 314 315 316 317 318
	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
319
		 */
320 321 322 323 324 325 326 327 328 329 330 331
		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
332 333
	}
	    
334 335 336 337
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
338 339
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
340 341
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
342 343 344 345 346 347 348 349
			if (err)
				goto out;
		}

		/*
		 * If we are changing the size of the file, then
		 * we need to break all leases.
		 */
350 351 352 353
		host_err = break_lease(inode, FMODE_WRITE | O_NONBLOCK);
		if (host_err == -EWOULDBLOCK)
			host_err = -ETIMEDOUT;
		if (host_err) /* ENOMEM or EWOULDBLOCK */
L
Linus Torvalds 已提交
354 355
			goto out_nfserr;

356 357
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
358 359 360
			goto out_nfserr;

		size_change = 1;
361 362
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
363 364 365
			put_write_access(inode);
			goto out_nfserr;
		}
366
		vfs_dq_init(inode);
L
Linus Torvalds 已提交
367 368
	}

369
	/* sanitize the mode change */
L
Linus Torvalds 已提交
370 371
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
372
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
373 374 375
	}

	/* Revoke setuid/setgid on chown */
376 377 378
	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))) {
379 380 381
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
382
			iap->ia_mode &= ~S_ISUID;
383 384 385 386 387
			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);
388
		}
L
Linus Torvalds 已提交
389 390 391 392 393 394 395 396 397
	}

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
		fh_lock(fhp);
398 399
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
400 401 402 403 404 405 406 407 408 409 410
		fh_unlock(fhp);
	}
	if (size_change)
		put_write_access(inode);
	if (!err)
		if (EX_ISSYNC(fhp->fh_export))
			write_inode_now(inode, 1);
out:
	return err;

out_nfserr:
411
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
412 413 414
	goto out;
}

415 416 417 418 419 420
#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;
421
	ssize_t ret;
422 423 424 425

	buflen = vfs_getxattr(dentry, key, NULL, 0);
	if (buflen <= 0)
		return buflen;
L
Linus Torvalds 已提交
426

427 428 429 430
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

431 432 433 434
	ret = vfs_getxattr(dentry, key, *buf, buflen);
	if (ret < 0)
		kfree(*buf);
	return ret;
435 436 437 438
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
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;
	}

459
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
460 461 462 463 464
out:
	kfree(buf);
	return error;
}

465
__be32
L
Linus Torvalds 已提交
466 467 468
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
469 470
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
471 472 473 474 475 476
	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
477
	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
478
	if (error)
479
		return error;
L
Linus Torvalds 已提交
480 481 482 483 484 485

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

486 487
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
488
		return nfserr_attrnotsupp;
489
	} else if (host_error < 0)
L
Linus Torvalds 已提交
490 491
		goto out_nfserr;

492 493
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
494
		goto out_release;
L
Linus Torvalds 已提交
495

496
	if (S_ISDIR(inode->i_mode))
497
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
498

499
out_release:
L
Linus Torvalds 已提交
500 501 502
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
503
	if (host_error == -EOPNOTSUPP)
504
		return nfserr_attrnotsupp;
505
	else
506
		return nfserrno(host_error);
L
Linus Torvalds 已提交
507 508 509 510 511
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
512
	void *buf = NULL;
L
Linus Torvalds 已提交
513
	struct posix_acl *pacl = NULL;
514
	int buflen;
L
Linus Torvalds 已提交
515

516 517 518 519 520
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
521 522 523 524 525 526 527 528 529 530 531 532 533 534

	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;

535
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
536 537 538 539 540 541 542 543 544
	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)) {
545
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
		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;
}

#endif /* defined(CONFIG_NFS_V4) */

#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
578 579 580 581
    {	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
582 583 584 585 586

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
587 588 589 590 591
    {	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
592 593 594 595 596 597 598 599 600 601 602 603

    {	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
604 605 606 607
    {	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
608 609 610 611

    {	0,			0				}
};

612
__be32
L
Linus Torvalds 已提交
613 614 615 616 617 618
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;
619
	__be32			error;
L
Linus Torvalds 已提交
620

M
Miklos Szeredi 已提交
621
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
	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) {
639
			__be32 err2;
L
Linus Torvalds 已提交
640 641 642

			sresult |= map->access;

643
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
			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
 */
678
__be32
L
Linus Torvalds 已提交
679 680 681
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
			int access, struct file **filp)
{
682
	const struct cred *cred = current_cred();
L
Linus Torvalds 已提交
683 684
	struct dentry	*dentry;
	struct inode	*inode;
685 686 687
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
688 689 690 691 692 693

	/*
	 * 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
694
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
695 696 697 698 699 700 701 702 703 704
	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
705
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
706
		goto out;
J
J. Bruce Fields 已提交
707 708 709 710 711 712
	/*
	 * 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
713 714 715 716 717 718 719 720 721
		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.
	 */
M
Miklos Szeredi 已提交
722
	host_err = break_lease(inode, O_NONBLOCK | ((access & NFSD_MAY_WRITE) ? FMODE_WRITE : 0));
723 724 725
	if (host_err == -EWOULDBLOCK)
		host_err = -ETIMEDOUT;
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
726 727
		goto out_nfserr;

M
Miklos Szeredi 已提交
728 729
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
730 731 732
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
733

734
		vfs_dq_init(inode);
L
Linus Torvalds 已提交
735
	}
736
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
737
			    flags, cred);
L
Linus Torvalds 已提交
738
	if (IS_ERR(*filp))
739
		host_err = PTR_ERR(*filp);
M
Mimi Zohar 已提交
740 741
	else
		ima_counts_get(*filp);
L
Linus Torvalds 已提交
742
out_nfserr:
743
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
744 745 746 747 748 749 750 751 752 753 754 755 756
out:
	return err;
}

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

757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777
/*
 * Sync a file
 * As this calls fsync (not fdatasync) there is no need for a write_inode
 * after it.
 */
static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
			      const struct file_operations *fop)
{
	struct inode *inode = dp->d_inode;
	int (*fsync) (struct file *, struct dentry *, int);
	int err;

	err = filemap_fdatawrite(inode->i_mapping);
	if (err == 0 && fop && (fsync = fop->fsync))
		err = fsync(filp, dp, 0);
	if (err == 0)
		err = filemap_fdatawait(inode->i_mapping);

	return err;
}

778
static int
L
Linus Torvalds 已提交
779 780
nfsd_sync(struct file *filp)
{
781 782 783 784 785 786 787 788
        int err;
	struct inode *inode = filp->f_path.dentry->d_inode;
	dprintk("nfsd: sync file %s\n", filp->f_path.dentry->d_name.name);
	mutex_lock(&inode->i_mutex);
	err=nfsd_dosync(filp, filp->f_path.dentry, filp->f_op);
	mutex_unlock(&inode->i_mutex);

	return err;
L
Linus Torvalds 已提交
789 790
}

791
int
792
nfsd_sync_dir(struct dentry *dp)
L
Linus Torvalds 已提交
793
{
794
	return nfsd_dosync(NULL, dp, dp->d_inode->i_fop);
L
Linus Torvalds 已提交
795 796 797 798 799 800 801 802 803 804 805 806
}

/*
 * 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;
807 808 809 810 811
	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
812

813 814
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
815 816 817 818 819 820 821 822
		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) {	
823
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
824 825 826 827 828 829 830
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
831
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
832
found:
833
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
834
		*rap = ra->p_next;
835 836
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
837 838 839
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
840
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
841 842 843 844
	return ra;
}

/*
845 846 847
 * 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
848 849
 */
static int
850 851
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
852
{
853
	struct svc_rqst *rqstp = sd->u.data;
854
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
855 856 857
	struct page *page = buf->page;
	size_t size;
	int ret;
L
Linus Torvalds 已提交
858

859
	ret = buf->ops->confirm(pipe, buf);
860 861 862 863
	if (unlikely(ret))
		return ret;

	size = sd->len;
L
Linus Torvalds 已提交
864 865 866

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
867 868 869
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
870
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
871
		rqstp->rq_res.page_len = size;
872
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
873
		get_page(page);
874 875
		if (*pp)
			put_page(*pp);
876 877
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
878
		rqstp->rq_res.page_len += size;
879
	} else
L
Linus Torvalds 已提交
880 881 882 883 884
		rqstp->rq_res.page_len += size;

	return size;
}

885 886 887 888 889 890
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

891 892 893 894 895 896 897 898 899
static inline int svc_msnfs(struct svc_fh *ffhp)
{
#ifdef MSNFS
	return (ffhp->fh_export->ex_flags & NFSEXP_MSNFS);
#else
	return 0;
#endif
}

900
static __be32
L
Linus Torvalds 已提交
901 902 903 904 905 906
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;
	struct raparms	*ra;
	mm_segment_t	oldfs;
907 908
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
909 910

	err = nfserr_perm;
911
	inode = file->f_path.dentry->d_inode;
912 913

	if (svc_msnfs(fhp) && !lock_may_read(inode, offset, *count))
L
Linus Torvalds 已提交
914 915 916 917 918 919 920 921
		goto out;

	/* 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;

922 923 924 925 926 927 928 929
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

930
		rqstp->rq_resused = 1;
931
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
932 933 934
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
935
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
936 937 938 939 940
		set_fs(oldfs);
	}

	/* Write back readahead params */
	if (ra) {
941 942
		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
		spin_lock(&rab->pb_lock);
L
Linus Torvalds 已提交
943 944 945
		ra->p_ra = file->f_ra;
		ra->p_set = 1;
		ra->p_count--;
946
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
947 948
	}

949 950 951
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
952
		err = 0;
953
		fsnotify_access(file->f_path.dentry);
L
Linus Torvalds 已提交
954
	} else 
955
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
956 957 958 959
out:
	return err;
}

960 961 962
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
963
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
964

965
	mutex_lock(&dentry->d_inode->i_mutex);
966
	notify_change(dentry, &ia);
967
	mutex_unlock(&dentry->d_inode->i_mutex);
968 969
}

970
static __be32
L
Linus Torvalds 已提交
971 972
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
973
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
974 975 976 977 978
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
979 980
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
981 982
	int			stable = *stablep;

983
#ifdef MSNFS
L
Linus Torvalds 已提交
984 985 986
	err = nfserr_perm;

	if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
987
		(!lock_may_write(file->f_path.dentry->d_inode, offset, *cnt)))
L
Linus Torvalds 已提交
988 989 990
		goto out;
#endif

991
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
992 993 994 995 996 997 998 999 1000 1001 1002 1003
	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().
	 * When gathered writes have been configured for this volume,
	 * flushing the data to disk is handled separately below.
	 */

1004
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
1005 1006 1007 1008 1009 1010
	       stable = 2;
	       *stablep = 2; /* FILE_SYNC */
	}

	if (!EX_ISSYNC(exp))
		stable = 0;
J
Jonathan Corbet 已提交
1011 1012
	if (stable && !EX_WGATHER(exp)) {
		spin_lock(&file->f_lock);
L
Linus Torvalds 已提交
1013
		file->f_flags |= O_SYNC;
J
Jonathan Corbet 已提交
1014 1015
		spin_unlock(&file->f_lock);
	}
L
Linus Torvalds 已提交
1016 1017 1018

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
1019
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1020
	set_fs(oldfs);
1021
	if (host_err >= 0) {
1022
		*cnt = host_err;
1023
		nfsdstats.io_write += host_err;
1024
		fsnotify_modify(file->f_path.dentry);
L
Linus Torvalds 已提交
1025 1026 1027
	}

	/* clear setuid/setgid flag after write */
1028
	if (host_err >= 0 && (inode->i_mode & (S_ISUID | S_ISGID)))
1029
		kill_suid(dentry);
L
Linus Torvalds 已提交
1030

1031
	if (host_err >= 0 && stable) {
L
Linus Torvalds 已提交
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049
		static ino_t	last_ino;
		static dev_t	last_dev;

		/*
		 * 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:-)
		 */
		if (EX_WGATHER(exp)) {
			if (atomic_read(&inode->i_writecount) > 1
			    || (last_ino == inode->i_ino && last_dev == inode->i_sb->s_dev)) {
1050
				dprintk("nfsd: write defer %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1051
				msleep(10);
1052
				dprintk("nfsd: write resume %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1053 1054 1055
			}

			if (inode->i_state & I_DIRTY) {
1056
				dprintk("nfsd: write sync %d\n", task_pid_nr(current));
1057
				host_err=nfsd_sync(file);
L
Linus Torvalds 已提交
1058 1059 1060 1061 1062 1063 1064 1065 1066
			}
#if 0
			wake_up(&inode->i_wait);
#endif
		}
		last_ino = inode->i_ino;
		last_dev = inode->i_sb->s_dev;
	}

1067
	dprintk("nfsd: write complete host_err=%d\n", host_err);
1068
	if (host_err >= 0)
L
Linus Torvalds 已提交
1069
		err = 0;
1070
	else
1071
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1072 1073 1074 1075 1076 1077 1078 1079 1080
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
 */
1081
__be32
L
Linus Torvalds 已提交
1082 1083 1084 1085
nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1086
	__be32		err;
L
Linus Torvalds 已提交
1087 1088

	if (file) {
1089
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1090
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1091 1092 1093 1094
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
	} else {
M
Miklos Szeredi 已提交
1095
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
		nfsd_close(file);
	}
out:
	return err;
}

/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1110
__be32
L
Linus Torvalds 已提交
1111
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1112
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1113 1114
		int *stablep)
{
1115
	__be32			err = 0;
L
Linus Torvalds 已提交
1116 1117

	if (file) {
1118
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1119
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1120 1121 1122 1123 1124
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1125
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
		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.
 * Strictly speaking, we could sync just the indicated file region here,
 * but there's currently no way we can ask the VFS to do so.
 *
 * 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.
 */
1147
__be32
L
Linus Torvalds 已提交
1148 1149 1150 1151
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1152
	__be32		err;
L
Linus Torvalds 已提交
1153 1154 1155 1156

	if ((u64)count > ~(u64)offset)
		return nfserr_inval;

M
Miklos Szeredi 已提交
1157 1158
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
	if (err)
L
Linus Torvalds 已提交
1159 1160 1161
		return err;
	if (EX_ISSYNC(fhp->fh_export)) {
		if (file->f_op && file->f_op->fsync) {
1162
			err = nfserrno(nfsd_sync(file));
L
Linus Torvalds 已提交
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
		} else {
			err = nfserr_notsupp;
		}
	}

	nfsd_close(file);
	return err;
}
#endif /* CONFIG_NFSD_V3 */

A
Adrian Bunk 已提交
1173
static __be32
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
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:
	 */
1186
	if (current_fsuid() != 0)
1187 1188 1189 1190 1191 1192
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
/* 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
1208 1209 1210 1211 1212 1213 1214 1215
/*
 * 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
 */
1216
__be32
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222
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;
1223
	__be32		err;
1224
	__be32		err2;
1225
	int		host_err;
L
Linus Torvalds 已提交
1226 1227 1228 1229 1230 1231 1232 1233

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

M
Miklos Szeredi 已提交
1234
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1235 1236 1237 1238 1239 1240 1241
	if (err)
		goto out;

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

	err = nfserr_notdir;
A
Al Viro 已提交
1242
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1243 1244 1245 1246 1247 1248 1249
		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
1250
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1251
		dchild = lookup_one_len(fname, dentry, flen);
1252
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
		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
1267
			err = nfserr_io;
L
Linus Torvalds 已提交
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
			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;

1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
	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
1296 1297 1298
	/*
	 * Get the dir op function pointer.
	 */
1299
	err = 0;
L
Linus Torvalds 已提交
1300 1301
	switch (type) {
	case S_IFREG:
1302
		host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1303 1304
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
L
Linus Torvalds 已提交
1305 1306
		break;
	case S_IFDIR:
1307
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1308 1309 1310 1311 1312
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1313
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1314 1315
		break;
	}
1316 1317
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1318
		goto out_nfserr;
1319
	}
L
Linus Torvalds 已提交
1320 1321

	if (EX_ISSYNC(fhp->fh_export)) {
1322
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1323 1324 1325
		write_inode_now(dchild->d_inode, 1);
	}

1326 1327 1328
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1329
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
	/*
	 * 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:
1341
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1342 1343 1344 1345 1346 1347 1348
	goto out;
}

#ifdef CONFIG_NFSD_V3
/*
 * NFSv3 version of nfsd_create
 */
1349
__be32
L
Linus Torvalds 已提交
1350 1351 1352
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,
1353
	        int *truncp, int *created)
L
Linus Torvalds 已提交
1354 1355 1356
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1357
	__be32		err;
1358
	__be32		err2;
1359
	int		host_err;
L
Linus Torvalds 已提交
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
	__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
1370
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1371 1372 1373 1374 1375 1376 1377 1378 1379
	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
1380
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1381
		goto out;
P
Peter Zijlstra 已提交
1382
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1383 1384 1385 1386 1387

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1388
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1389 1390 1391 1392 1393 1394 1395 1396
	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) {
1397
		/* solaris7 gets confused (bugid 4218508) if these have
1398 1399 1400 1401
		 * 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
1402 1403 1404 1405 1406
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1407 1408 1409
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
	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;
		}
1441
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1442 1443 1444
		goto out;
	}

1445
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1446 1447
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1448
		goto out_nfserr;
1449
	}
1450 1451
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1452 1453

	if (EX_ISSYNC(fhp->fh_export)) {
1454
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1455 1456 1457
		/* setattr will sync the child (or not) */
	}

1458 1459
	nfsd_check_ignore_resizing(iap);

L
Linus Torvalds 已提交
1460
	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1461
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1462
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1463
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1464 1465 1466 1467 1468 1469 1470 1471
		/* 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:
1472 1473 1474
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1475

1476
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
1477 1478 1479 1480 1481
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487 1488 1489

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1490
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499
	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
 */
1500
__be32
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1506 1507
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1508

M
Miklos Szeredi 已提交
1509
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514 1515 1516
	if (err)
		goto out;

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

	err = nfserr_inval;
A
Al Viro 已提交
1517
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1518 1519
		goto out;

1520
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1521 1522 1523 1524 1525
	/* 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);
1526
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1527 1528
	set_fs(oldfs);

1529
	if (host_err < 0)
L
Linus Torvalds 已提交
1530
		goto out_nfserr;
1531
	*lenp = host_err;
L
Linus Torvalds 已提交
1532 1533 1534 1535 1536
	err = 0;
out:
	return err;

out_nfserr:
1537
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1538 1539 1540 1541 1542 1543 1544
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1545
__be32
L
Linus Torvalds 已提交
1546 1547 1548 1549 1550 1551 1552
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;
1553 1554
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1555 1556 1557 1558 1559 1560 1561 1562

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

M
Miklos Szeredi 已提交
1563
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1564 1565 1566 1567 1568
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1569
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1570 1571 1572
	if (IS_ERR(dnew))
		goto out_nfserr;

1573 1574 1575 1576
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1577 1578 1579
	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1580
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1581 1582 1583
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1584
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1585 1586 1587
			kfree(path_alloced);
		}
	} else
1588
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
L
Linus Torvalds 已提交
1589

1590
	if (!host_err) {
L
Linus Torvalds 已提交
1591
		if (EX_ISSYNC(fhp->fh_export))
1592 1593 1594
			host_err = nfsd_sync_dir(dentry);
	}
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1595 1596
	fh_unlock(fhp);

1597 1598
	mnt_drop_write(fhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1599 1600 1601 1602 1603 1604 1605
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1606
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1607 1608 1609 1610 1611 1612 1613
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1614
__be32
L
Linus Torvalds 已提交
1615 1616 1617 1618 1619
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
	struct inode	*dirp, *dest;
1620 1621
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1622

M
Miklos Szeredi 已提交
1623
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1624 1625
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1626
	err = fh_verify(rqstp, tfhp, -S_IFDIR, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1627 1628 1629 1630 1631 1632 1633 1634 1635 1636
	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
1637
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1638 1639 1640 1641
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1642
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647 1648
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;
	dest = dold->d_inode;

1649 1650 1651 1652 1653
	host_err = mnt_want_write(tfhp->fh_export->ex_path.mnt);
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1654 1655
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
L
Linus Torvalds 已提交
1656
		if (EX_ISSYNC(ffhp->fh_export)) {
1657
			err = nfserrno(nfsd_sync_dir(ddir));
L
Linus Torvalds 已提交
1658 1659
			write_inode_now(dest, 1);
		}
1660
		err = 0;
L
Linus Torvalds 已提交
1661
	} else {
1662
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1663 1664
			err = nfserr_acces;
		else
1665
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1666
	}
1667 1668
	mnt_drop_write(tfhp->fh_export->ex_path.mnt);
out_dput:
L
Linus Torvalds 已提交
1669
	dput(dnew);
1670 1671
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1672 1673 1674 1675
out:
	return err;

out_nfserr:
1676
	err = nfserrno(host_err);
1677
	goto out_unlock;
L
Linus Torvalds 已提交
1678 1679 1680 1681 1682 1683
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1684
__be32
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689
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;
1690 1691
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1692

M
Miklos Szeredi 已提交
1693
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1694 1695
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1696
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
	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;
1707
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
		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);
1722
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1723 1724 1725
	if (IS_ERR(odentry))
		goto out_nfserr;

1726
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1727 1728
	if (!odentry->d_inode)
		goto out_dput_old;
1729
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1730 1731 1732 1733
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1734
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1735 1736
	if (IS_ERR(ndentry))
		goto out_dput_old;
1737
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1738 1739 1740
	if (ndentry == trap)
		goto out_dput_new;

1741
	if (svc_msnfs(ffhp) &&
L
Linus Torvalds 已提交
1742 1743
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
1744
			host_err = -EPERM;
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
			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;

1755 1756 1757 1758 1759
	host_err = vfs_rename(fdir, odentry, tdir, ndentry);
	if (!host_err && EX_ISSYNC(tfhp->fh_export)) {
		host_err = nfsd_sync_dir(tdentry);
		if (!host_err)
			host_err = nfsd_sync_dir(fdentry);
L
Linus Torvalds 已提交
1760 1761
	}

1762 1763
	mnt_drop_write(ffhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1764 1765 1766 1767 1768
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1769
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787

	/* 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
 */
1788
__be32
L
Linus Torvalds 已提交
1789 1790 1791 1792 1793
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1794 1795
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1796 1797 1798 1799

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1800
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1801 1802 1803
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1804
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1805 1806 1807 1808
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1809
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821
	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;

1822 1823 1824 1825
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1826 1827 1828 1829
	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
1830
			host_err = -EPERM;
L
Linus Torvalds 已提交
1831 1832
		} else
#endif
1833
		host_err = vfs_unlink(dirp, rdentry);
L
Linus Torvalds 已提交
1834
	} else { /* It's RMDIR */
1835
		host_err = vfs_rmdir(dirp, rdentry);
L
Linus Torvalds 已提交
1836 1837 1838 1839
	}

	dput(rdentry);

1840
	if (host_err)
1841
		goto out_drop;
1842 1843
	if (EX_ISSYNC(fhp->fh_export))
		host_err = nfsd_sync_dir(dentry);
L
Linus Torvalds 已提交
1844

1845 1846
out_drop:
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1847
out_nfserr:
1848
	err = nfserrno(host_err);
1849 1850
out:
	return err;
L
Linus Torvalds 已提交
1851 1852
}

1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
/*
 * 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;
1871
	int		full;
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881
};

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));
1882 1883
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1884
		return -EINVAL;
1885
	}
1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896

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

	return 0;
}

1897 1898
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1899
{
1900 1901
	struct readdir_data buf;
	struct buffered_dirent *de;
1902
	int host_err;
1903 1904
	int size;
	loff_t offset;
1905

1906 1907
	buf.dirent = (void *)__get_free_page(GFP_KERNEL);
	if (!buf.dirent)
1908
		return nfserrno(-ENOMEM);
1909 1910

	offset = *offsetp;
1911

1912
	while (1) {
1913
		struct inode *dir_inode = file->f_path.dentry->d_inode;
1914 1915
		unsigned int reclen;

1916
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1917
		buf.used = 0;
1918
		buf.full = 0;
1919 1920

		host_err = vfs_readdir(file, nfsd_buffered_filldir, &buf);
1921 1922 1923 1924
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1925 1926 1927 1928 1929 1930 1931
			break;

		size = buf.used;

		if (!size)
			break;

1932 1933 1934 1935 1936 1937 1938 1939 1940
		/*
		 * 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;

1941 1942 1943 1944 1945 1946
		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))
1947
				break;
1948 1949

			if (cdp->err != nfs_ok)
1950
				break;
1951 1952 1953 1954 1955 1956

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1957 1958 1959 1960
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1961
		offset = vfs_llseek(file, 0, SEEK_CUR);
1962 1963 1964
	}

	free_page((unsigned long)(buf.dirent));
1965 1966 1967

	if (host_err)
		return nfserrno(host_err);
1968 1969 1970

	*offsetp = offset;
	return cdp->err;
1971 1972
}

L
Linus Torvalds 已提交
1973 1974 1975 1976
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1977
__be32
L
Linus Torvalds 已提交
1978
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1979
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1980
{
1981
	__be32		err;
L
Linus Torvalds 已提交
1982 1983 1984
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
1985
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1986 1987 1988 1989 1990 1991 1992 1993 1994
	if (err)
		goto out;

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

1995
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

	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
 */
2009
__be32
2010
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
2011
{
2012
	__be32 err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
2013
	if (!err && vfs_statfs(fhp->fh_dentry,stat))
L
Linus Torvalds 已提交
2014 2015 2016 2017
		err = nfserr_io;
	return err;
}

J
J. Bruce Fields 已提交
2018
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
2019
{
J
J. Bruce Fields 已提交
2020
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
2021 2022
}

L
Linus Torvalds 已提交
2023 2024 2025
/*
 * Check for a user's access permissions to this inode.
 */
2026
__be32
2027 2028
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
2029 2030
{
	struct inode	*inode = dentry->d_inode;
M
Mimi Zohar 已提交
2031
	struct path	path;
L
Linus Torvalds 已提交
2032 2033
	int		err;

M
Miklos Szeredi 已提交
2034
	if (acc == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
2035 2036 2037 2038
		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
2039 2040 2041 2042 2043 2044 2045
		(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
2046 2047 2048
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2049
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2050
	dprintk("      owner %d/%d user %d/%d\n",
2051
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2052 2053 2054 2055 2056 2057
#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
2058 2059
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2060 2061
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2062
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2063
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2064 2065
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2066
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2067 2068
		return nfserr_perm;

M
Miklos Szeredi 已提交
2069
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2070 2071 2072 2073 2074 2075 2076
		/* 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
2077
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092
	}
	/*
	 * 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
2093
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2094
	    inode->i_uid == current_fsuid())
L
Linus Torvalds 已提交
2095 2096
		return 0;

M
Miklos Szeredi 已提交
2097
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2098
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2099 2100 2101

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
M
Miklos Szeredi 已提交
2102
	    acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE))
2103
		err = inode_permission(inode, MAY_EXEC);
M
Mimi Zohar 已提交
2104 2105
	if (err)
		goto nfsd_out;
L
Linus Torvalds 已提交
2106

M
Mimi Zohar 已提交
2107 2108 2109 2110 2111 2112 2113 2114
	/* Do integrity (permission) checking now, but defer incrementing
	 * IMA counts to the actual file open.
	 */
	path.mnt = exp->ex_path.mnt;
	path.dentry = dentry;
	err = ima_path_check(&path, acc & (MAY_READ | MAY_WRITE | MAY_EXEC),
			     IMA_COUNT_LEAVE);
nfsd_out:
L
Linus Torvalds 已提交
2115 2116 2117 2118 2119 2120
	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2121 2122 2123
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2124
	dprintk("nfsd: freeing readahead buffers.\n");
2125 2126 2127 2128 2129 2130 2131 2132 2133 2134

	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
2135 2136 2137 2138 2139 2140 2141 2142
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2143 2144
	int	j = 0;
	int	nperbucket;
2145
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2146

2147

2148
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2149
		return 0;
2150 2151 2152 2153
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2154 2155

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2156 2157

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2158
		spin_lock_init(&raparm_hash[i].pb_lock);
2159 2160 2161 2162 2163 2164 2165 2166 2167

		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;
2168 2169
	}

L
Linus Torvalds 已提交
2170 2171
	nfsdstats.ra_size = cache_size;
	return 0;
2172 2173 2174 2175 2176

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2177
}
2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188

#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;

2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
	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:
2200 2201 2202
		return ERR_PTR(-EOPNOTSUPP);
	}

2203 2204 2205
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220

	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
2221
	if (!IS_POSIXACL(inode) ||
2222 2223 2224 2225
	    !inode->i_op->setxattr || !inode->i_op->removexattr)
		return -EOPNOTSUPP;
	switch(type) {
		case ACL_TYPE_ACCESS:
2226
			name = POSIX_ACL_XATTR_ACCESS;
2227 2228
			break;
		case ACL_TYPE_DEFAULT:
2229
			name = POSIX_ACL_XATTR_DEFAULT;
2230 2231 2232 2233 2234 2235
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2236
		size = posix_acl_xattr_size(acl->a_count);
2237 2238 2239
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2240 2241
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2242
			goto getout;
2243
		size = error;
2244 2245 2246
	} else
		size = 0;

2247 2248 2249
	error = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (error)
		goto getout;
2250
	if (size)
2251
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2252 2253 2254 2255
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2256
			error = vfs_removexattr(fhp->fh_dentry, name);
2257 2258 2259 2260
			if (error == -ENODATA)
				error = 0;
		}
	}
2261
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
2262 2263 2264 2265 2266 2267

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