vfs.c 50.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * File operations used by nfsd. Some of these have been ripped from
 * other parts of the kernel because they weren't exported, others
 * are partial duplicates with added or changed functionality.
 *
 * Note that several functions dget() the dentry upon which they want
 * to act, most notably those that create directory entries. Response
 * dentry's are dput()'d if necessary in the release callback.
 * So if you notice code paths that apparently fail to dput() the
 * dentry, don't worry--they have been taken care of.
 *
 * Copyright (C) 1995-1999 Olaf Kirch <okir@monad.swb.de>
 * Zerocpy NFS support (C) 2002 Hirokazu Takahashi <taka@valinux.co.jp>
 */

#include <linux/fs.h>
#include <linux/file.h>
18
#include <linux/splice.h>
L
Linus Torvalds 已提交
19 20 21
#include <linux/fcntl.h>
#include <linux/namei.h>
#include <linux/delay.h>
R
Robert Love 已提交
22
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
23 24
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
25 26
#include <linux/jhash.h>
#include <linux/ima.h>
27
#include <linux/slab.h>
28
#include <asm/uaccess.h>
29 30
#include <linux/exportfs.h>
#include <linux/writeback.h>
31
#include <linux/security.h>
32 33 34 35 36

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

37
#ifdef CONFIG_NFSD_V4
38 39
#include "acl.h"
#include "idmap.h"
L
Linus Torvalds 已提交
40 41
#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

91
	err = follow_down(&path);
92 93
	if (err < 0)
		goto out;
L
Linus Torvalds 已提交
94

A
Al Viro 已提交
95
	exp2 = rqst_exp_get_by_name(rqstp, &path);
L
Linus Torvalds 已提交
96
	if (IS_ERR(exp2)) {
97 98 99 100 101 102 103 104 105 106
		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
107
		path_put(&path);
L
Linus Torvalds 已提交
108 109
		goto out;
	}
110 111
	if (nfsd_v4client(rqstp) ||
		(exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
L
Linus Torvalds 已提交
112
		/* successfully crossed mount point */
A
Al Viro 已提交
113
		/*
A
Al Viro 已提交
114 115 116 117
		 * 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
118
		 */
A
Al Viro 已提交
119 120
		*dpp = path.dentry;
		path.dentry = dentry;
A
Al Viro 已提交
121
		*expp = exp2;
A
Al Viro 已提交
122
		exp2 = exp;
L
Linus Torvalds 已提交
123
	}
A
Al Viro 已提交
124 125
	path_put(&path);
	exp_put(exp2);
L
Linus Torvalds 已提交
126 127 128 129
out:
	return err;
}

130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
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;
}

164 165 166 167
/*
 * For nfsd purposes, we treat V4ROOT exports as though there was an
 * export at *every* directory.
 */
168
int nfsd_mountpoint(struct dentry *dentry, struct svc_export *exp)
169 170 171
{
	if (d_mountpoint(dentry))
		return 1;
172 173
	if (nfsd4_is_junction(dentry))
		return 1;
174 175 176 177 178
	if (!(exp->ex_flags & NFSEXP_V4ROOT))
		return 0;
	return dentry->d_inode != NULL;
}

179
__be32
180
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
181
		   const char *name, unsigned int len,
182
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
183 184 185 186
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
187
	int			host_err;
L
Linus Torvalds 已提交
188 189 190 191 192 193 194 195 196 197 198

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

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

	/* Lookup the name, but don't follow links */
	if (isdotent(name, len)) {
		if (len==1)
			dentry = dget(dparent);
199
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
200
			dentry = dget_parent(dparent);
201
		else if (!EX_NOHIDE(exp) && !nfsd_v4client(rqstp))
L
Linus Torvalds 已提交
202 203 204
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
205 206
			host_err = nfsd_lookup_parent(rqstp, dparent, &exp, &dentry);
			if (host_err)
L
Linus Torvalds 已提交
207 208 209
				goto out_nfserr;
		}
	} else {
210 211 212 213 214 215
		/*
		 * In the nfsd4_open() case, this may be held across
		 * subsequent open and delegation acquisition which may
		 * need to take the child's i_mutex:
		 */
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
216
		dentry = lookup_one_len(name, dparent, len);
217
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
218 219 220 221 222
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
223
		if (nfsd_mountpoint(dentry, exp)) {
224
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
225 226 227 228 229
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
	*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,
253
				unsigned int len, struct svc_fh *resfh)
254 255 256 257 258
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

259 260 261
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
	if (err)
		return err;
262 263 264
	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
265 266 267
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
268 269 270 271 272 273 274
	/*
	 * 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;
275
out:
L
Linus Torvalds 已提交
276 277 278 279 280
	dput(dentry);
	exp_put(exp);
	return err;
}

281 282 283 284 285 286 287 288 289 290 291 292
/*
 * 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
293 294 295
	if (export_ops->commit_metadata)
		return export_ops->commit_metadata(inode);
	return sync_inode_metadata(inode, 1);
296
}
297

L
Linus Torvalds 已提交
298
/*
C
Christoph Hellwig 已提交
299 300
 * Go over the attributes and take care of the small differences between
 * NFS semantics and what Linux expects.
L
Linus Torvalds 已提交
301
 */
C
Christoph Hellwig 已提交
302 303
static void
nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
L
Linus Torvalds 已提交
304
{
305 306
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
307 308 309 310 311 312 313
	 * 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
C
Christoph Hellwig 已提交
314
	 * it is not an interface that we should be using.
L
Linus Torvalds 已提交
315 316 317
	 */
#define BOTH_TIME_SET (ATTR_ATIME_SET | ATTR_MTIME_SET)
#define	MAX_TOUCH_TIME_ERROR (30*60)
318 319 320 321 322 323 324 325
	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
326
		 */
327 328 329 330 331 332 333 334 335 336 337 338
		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
339 340
	}

341
	/* sanitize the mode change */
L
Linus Torvalds 已提交
342 343
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
344
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
345 346 347
	}

	/* Revoke setuid/setgid on chown */
348
	if (!S_ISDIR(inode->i_mode) &&
349
	    ((iap->ia_valid & ATTR_UID) || (iap->ia_valid & ATTR_GID))) {
350 351 352
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
353
			iap->ia_mode &= ~S_ISUID;
354 355 356 357 358
			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);
359
		}
L
Linus Torvalds 已提交
360
	}
C
Christoph Hellwig 已提交
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
}

static __be32
nfsd_get_write_access(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct iattr *iap)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	int host_err;

	if (iap->ia_size < inode->i_size) {
		__be32 err;

		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
				NFSD_MAY_TRUNC | NFSD_MAY_OWNER_OVERRIDE);
		if (err)
			return err;
	}

	host_err = get_write_access(inode);
	if (host_err)
		goto out_nfserrno;

	host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
	if (host_err)
		goto out_put_write_access;
	return 0;

out_put_write_access:
	put_write_access(inode);
out_nfserrno:
	return nfserrno(host_err);
}

/*
 * Set various file attributes.  After this call fhp needs an fh_put.
 */
__be32
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;
	int		accmode = NFSD_MAY_SATTR;
	umode_t		ftype = 0;
	__be32		err;
	int		host_err;
407
	bool		get_write_count;
C
Christoph Hellwig 已提交
408 409 410 411 412 413 414
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

415 416 417
	/* Callers that do fh_verify should do the fh_want_write: */
	get_write_count = !fhp->fh_dentry;

C
Christoph Hellwig 已提交
418 419 420 421
	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
	if (err)
		goto out;
422 423 424 425 426
	if (get_write_count) {
		host_err = fh_want_write(fhp);
		if (host_err)
			return nfserrno(host_err);
	}
C
Christoph Hellwig 已提交
427 428 429 430 431 432 433 434 435 436

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

	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;
L
Linus Torvalds 已提交
437

C
Christoph Hellwig 已提交
438 439 440 441 442 443 444 445 446 447 448 449
	nfsd_sanitize_attrs(inode, iap);

	/*
	 * The size case is special, it changes the file in addition to the
	 * attributes.
	 */
	if (iap->ia_valid & ATTR_SIZE) {
		err = nfsd_get_write_access(rqstp, fhp, iap);
		if (err)
			goto out;
		size_change = 1;
	}
L
Linus Torvalds 已提交
450 451 452

	iap->ia_valid |= ATTR_CTIME;

453 454 455
	if (check_guard && guardtime != inode->i_ctime.tv_sec) {
		err = nfserr_notsync;
		goto out_put_write_access;
L
Linus Torvalds 已提交
456
	}
457 458 459 460

	fh_lock(fhp);
	host_err = notify_change(dentry, iap, NULL);
	fh_unlock(fhp);
461
	err = nfserrno(host_err);
462 463

out_put_write_access:
L
Linus Torvalds 已提交
464 465 466
	if (size_change)
		put_write_access(inode);
	if (!err)
467
		commit_metadata(fhp);
L
Linus Torvalds 已提交
468 469 470 471
out:
	return err;
}

472
#if defined(CONFIG_NFSD_V4)
473 474 475 476 477 478 479 480 481 482 483 484 485
/*
 * NFS junction information is stored in an extended attribute.
 */
#define NFSD_JUNCTION_XATTR_NAME	XATTR_TRUSTED_PREFIX "junction.nfs"

/**
 * nfsd4_is_junction - Test if an object could be an NFS junction
 *
 * @dentry: object to test
 *
 * Returns 1 if "dentry" appears to contain NFS junction information.
 * Otherwise 0 is returned.
 */
486 487 488 489 490 491 492 493 494 495
int nfsd4_is_junction(struct dentry *dentry)
{
	struct inode *inode = dentry->d_inode;

	if (inode == NULL)
		return 0;
	if (inode->i_mode & S_IXUGO)
		return 0;
	if (!(inode->i_mode & S_ISVTX))
		return 0;
496
	if (vfs_getxattr(dentry, NFSD_JUNCTION_XATTR_NAME, NULL, 0) <= 0)
497 498 499
		return 0;
	return 1;
}
500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
#ifdef CONFIG_NFSD_V4_SECURITY_LABEL
__be32 nfsd4_set_nfs4_label(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct xdr_netobj *label)
{
	__be32 error;
	int host_error;
	struct dentry *dentry;

	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
	if (error)
		return error;

	dentry = fhp->fh_dentry;

	mutex_lock(&dentry->d_inode->i_mutex);
	host_error = security_inode_setsecctx(dentry, label->data, label->len);
	mutex_unlock(&dentry->d_inode->i_mutex);
	return nfserrno(host_error);
}
#else
__be32 nfsd4_set_nfs4_label(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct xdr_netobj *label)
{
	return nfserr_notsupp;
}
#endif

J
J. Bruce Fields 已提交
527
#endif /* defined(CONFIG_NFSD_V4) */
L
Linus Torvalds 已提交
528 529 530 531 532 533 534 535 536 537

#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
538 539 540 541
    {	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
542 543 544 545 546

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
547 548 549 550 551
    {	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
552 553 554 555 556 557 558 559 560 561 562 563

    {	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
564 565 566 567
    {	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
568 569 570 571

    {	0,			0				}
};

572
__be32
L
Linus Torvalds 已提交
573 574 575 576 577 578
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;
579
	__be32			error;
L
Linus Torvalds 已提交
580

M
Miklos Szeredi 已提交
581
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
	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) {
599
			__be32 err2;
L
Linus Torvalds 已提交
600 601 602

			sresult |= map->access;

603
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
			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 */

631 632 633
static int nfsd_open_break_lease(struct inode *inode, int access)
{
	unsigned int mode;
L
Linus Torvalds 已提交
634

635 636 637 638 639
	if (access & NFSD_MAY_NOT_BREAK_LEASE)
		return 0;
	mode = (access & NFSD_MAY_WRITE) ? O_WRONLY : O_RDONLY;
	return break_lease(inode, mode | O_NONBLOCK);
}
L
Linus Torvalds 已提交
640 641 642

/*
 * Open an existing file or directory.
643 644
 * The may_flags argument indicates the type of open (read/write/lock)
 * and additional flags.
L
Linus Torvalds 已提交
645 646
 * N.B. After this call fhp needs an fh_put
 */
647
__be32
A
Al Viro 已提交
648
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
649
			int may_flags, struct file **filp)
L
Linus Torvalds 已提交
650
{
651
	struct path	path;
L
Linus Torvalds 已提交
652
	struct inode	*inode;
653 654
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
655
	int		host_err = 0;
L
Linus Torvalds 已提交
656

657 658
	validate_process_creds();

L
Linus Torvalds 已提交
659 660 661 662
	/*
	 * 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.
663 664 665 666 667 668
	 *
	 * Arguably we should also allow the owner override for
	 * directories, but we never have and it doesn't seem to have
	 * caused anyone a problem.  If we were to change this, note
	 * also that our filldir callbacks would need a variant of
	 * lookup_one_len that doesn't check permissions.
L
Linus Torvalds 已提交
669
	 */
670 671 672
	if (type == S_IFREG)
		may_flags |= NFSD_MAY_OWNER_OVERRIDE;
	err = fh_verify(rqstp, fhp, type, may_flags);
L
Linus Torvalds 已提交
673 674 675
	if (err)
		goto out;

676 677 678
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
679 680 681 682 683

	/* Disallow write access to files with the append-only bit set
	 * or any access when mandatory locking enabled
	 */
	err = nfserr_perm;
684
	if (IS_APPEND(inode) && (may_flags & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
685
		goto out;
J
J. Bruce Fields 已提交
686 687 688 689 690 691
	/*
	 * 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
692 693 694 695 696
		goto out;

	if (!inode->i_fop)
		goto out;

697
	host_err = nfsd_open_break_lease(inode, may_flags);
698
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
699 700
		goto out_nfserr;

701 702
	if (may_flags & NFSD_MAY_WRITE) {
		if (may_flags & NFSD_MAY_READ)
703 704 705
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
706
	}
707
	*filp = dentry_open(&path, flags, current_cred());
708
	if (IS_ERR(*filp)) {
709
		host_err = PTR_ERR(*filp);
710 711
		*filp = NULL;
	} else {
712 713
		host_err = ima_file_check(*filp, may_flags);

714 715 716 717 718 719
		if (may_flags & NFSD_MAY_64BIT_COOKIE)
			(*filp)->f_mode |= FMODE_64BITHASH;
		else
			(*filp)->f_mode |= FMODE_32BITHASH;
	}

L
Linus Torvalds 已提交
720
out_nfserr:
721
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
722
out:
723
	validate_process_creds();
L
Linus Torvalds 已提交
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745
	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;
746 747 748 749 750
	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
751

752 753
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
754 755 756 757 758 759
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
760
	depth = nfsdstats.ra_size;
L
Linus Torvalds 已提交
761
	if (!frap) {	
762
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
763 764 765 766 767 768 769
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
770
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
771
found:
772
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
773
		*rap = ra->p_next;
774 775
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
776 777 778
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
779
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
780 781 782 783
	return ra;
}

/*
784 785 786
 * 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
787 788
 */
static int
789 790
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
791
{
792
	struct svc_rqst *rqstp = sd->u.data;
793
	struct page **pp = rqstp->rq_next_page;
794 795 796 797
	struct page *page = buf->page;
	size_t size;

	size = sd->len;
L
Linus Torvalds 已提交
798 799 800

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
801 802
		put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
803
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
804
		rqstp->rq_res.page_len = size;
805
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
806
		get_page(page);
807 808 809
		if (*rqstp->rq_next_page)
			put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
L
Linus Torvalds 已提交
810
		rqstp->rq_res.page_len += size;
811
	} else
L
Linus Torvalds 已提交
812 813 814 815 816
		rqstp->rq_res.page_len += size;

	return size;
}

817 818 819 820 821 822
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

823
__be32 nfsd_finish_read(struct file *file, unsigned long *count, int host_err)
L
Linus Torvalds 已提交
824
{
825 826 827
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
828
		fsnotify_access(file);
829
		return 0;
L
Linus Torvalds 已提交
830
	} else 
831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
		return nfserrno(host_err);
}

int nfsd_splice_read(struct svc_rqst *rqstp,
		     struct file *file, loff_t offset, unsigned long *count)
{
	struct splice_desc sd = {
		.len		= 0,
		.total_len	= *count,
		.pos		= offset,
		.u.data		= rqstp,
	};
	int host_err;

	rqstp->rq_next_page = rqstp->rq_respages + 1;
	host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
	return nfsd_finish_read(file, count, host_err);
}

int nfsd_readv(struct file *file, loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
	mm_segment_t oldfs;
	int host_err;

	oldfs = get_fs();
	set_fs(KERNEL_DS);
	host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
	set_fs(oldfs);
	return nfsd_finish_read(file, count, host_err);
}

static __be32
nfsd_vfs_read(struct svc_rqst *rqstp, struct file *file,
	      loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
{
	if (file->f_op->splice_read && rqstp->rq_splice_ok)
		return nfsd_splice_read(rqstp, file, offset, count);
	else
		return nfsd_readv(file, offset, vec, vlen, count);
L
Linus Torvalds 已提交
871 872
}

873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888
/*
 * 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)
{
A
Al Viro 已提交
889
	struct inode *inode = file_inode(file);
890 891 892 893 894 895 896 897 898 899 900 901 902
	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));
903
		err = vfs_fsync(file, 0);
904 905 906 907 908 909
	}
	last_ino = inode->i_ino;
	last_dev = inode->i_sb->s_dev;
	return err;
}

910
static __be32
L
Linus Torvalds 已提交
911 912
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
913
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
914 915 916 917 918
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
919 920
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
921
	int			stable = *stablep;
922
	int			use_wgather;
K
Kent Overstreet 已提交
923
	loff_t			pos = offset;
924 925 926 927 928 929 930 931 932 933
	unsigned int		pflags = current->flags;

	if (rqstp->rq_local)
		/*
		 * We want less throttling in balance_dirty_pages()
		 * and shrink_inactive_list() so that nfs to
		 * localhost doesn't cause nfsd to lock up due to all
		 * the client's dirty pages or its congested queue.
		 */
		current->flags |= PF_LESS_THROTTLE;
L
Linus Torvalds 已提交
934

935
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
936 937 938
	inode = dentry->d_inode;
	exp   = fhp->fh_export;

939
	use_wgather = (rqstp->rq_vers == 2) && EX_WGATHER(exp);
L
Linus Torvalds 已提交
940 941 942 943 944 945

	if (!EX_ISSYNC(exp))
		stable = 0;

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
K
Kent Overstreet 已提交
946
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &pos);
L
Linus Torvalds 已提交
947
	set_fs(oldfs);
948 949 950 951
	if (host_err < 0)
		goto out_nfserr;
	*cnt = host_err;
	nfsdstats.io_write += host_err;
952
	fsnotify_modify(file);
L
Linus Torvalds 已提交
953

954 955 956 957 958 959
	if (stable) {
		if (use_wgather)
			host_err = wait_for_concurrent_writes(file);
		else
			host_err = vfs_fsync_range(file, offset, offset+*cnt, 0);
	}
L
Linus Torvalds 已提交
960

961
out_nfserr:
962
	dprintk("nfsd: write complete host_err=%d\n", host_err);
963
	if (host_err >= 0)
L
Linus Torvalds 已提交
964
		err = 0;
965
	else
966
		err = nfserrno(host_err);
967 968
	if (rqstp->rq_local)
		tsk_restore_flags(current, pflags, PF_LESS_THROTTLE);
L
Linus Torvalds 已提交
969 970 971
	return err;
}

972 973
__be32 nfsd_get_tmp_read_open(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct file **file, struct raparms **ra)
974 975 976 977
{
	struct inode *inode;
	__be32 err;

978
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, file);
979 980 981
	if (err)
		return err;

982
	inode = file_inode(*file);
983 984

	/* Get readahead parameters */
985
	*ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);
986

987 988 989 990
	if (*ra && (*ra)->p_set)
		(*file)->f_ra = (*ra)->p_ra;
	return nfs_ok;
}
991

992 993
void nfsd_put_tmp_read_open(struct file *file, struct raparms *ra)
{
994 995 996 997 998 999 1000 1001 1002 1003
	/* 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);
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
}

/*
 * 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
 */
__be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
	loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
{
	struct file *file;
	struct raparms	*ra;
	__be32 err;

	err = nfsd_get_tmp_read_open(rqstp, fhp, &file, &ra);
	if (err)
		return err;

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

	nfsd_put_tmp_read_open(file, ra);

1026 1027 1028
	return err;
}

1029
/* As above, but use the provided file descriptor. */
1030
__be32
1031
nfsd_read_file(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
L
Linus Torvalds 已提交
1032 1033 1034
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1035
	__be32		err;
L
Linus Torvalds 已提交
1036 1037

	if (file) {
1038
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1039
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1040 1041
		if (err)
			goto out;
1042
		err = nfsd_vfs_read(rqstp, file, offset, vec, vlen, count);
1043 1044
	} 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
1045 1046 1047 1048 1049 1050 1051 1052 1053
out:
	return err;
}

/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1054
__be32
L
Linus Torvalds 已提交
1055
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1056
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1057 1058
		int *stablep)
{
1059
	__be32			err = 0;
L
Linus Torvalds 已提交
1060 1061

	if (file) {
1062
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1063
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1064 1065 1066 1067 1068
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1069
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
		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.
1085 1086 1087
 *
 * 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
1088 1089 1090 1091
 *
 * 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.
 */
1092
__be32
L
Linus Torvalds 已提交
1093 1094 1095 1096
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1097 1098
	loff_t		end = LLONG_MAX;
	__be32		err = nfserr_inval;
L
Linus Torvalds 已提交
1099

1100 1101 1102 1103 1104 1105 1106
	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
1107

1108 1109
	err = nfsd_open(rqstp, fhp, S_IFREG,
			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &file);
M
Miklos Szeredi 已提交
1110
	if (err)
1111
		goto out;
L
Linus Torvalds 已提交
1112
	if (EX_ISSYNC(fhp->fh_export)) {
1113
		int err2 = vfs_fsync_range(file, offset, end, 0);
1114 1115 1116 1117

		if (err2 != -EINVAL)
			err = nfserrno(err2);
		else
L
Linus Torvalds 已提交
1118 1119 1120 1121
			err = nfserr_notsupp;
	}

	nfsd_close(file);
1122
out:
L
Linus Torvalds 已提交
1123 1124 1125 1126
	return err;
}
#endif /* CONFIG_NFSD_V3 */

A
Adrian Bunk 已提交
1127
static __be32
1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139
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:
	 */
1140
	if (!uid_eq(current_fsuid(), GLOBAL_ROOT_UID))
1141 1142 1143 1144 1145 1146
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161
/* 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
1162 1163 1164 1165 1166 1167 1168 1169
/*
 * 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
 */
1170
__be32
L
Linus Torvalds 已提交
1171 1172 1173 1174 1175 1176
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;
1177
	__be32		err;
1178
	__be32		err2;
1179
	int		host_err;
L
Linus Torvalds 已提交
1180 1181 1182 1183 1184 1185 1186 1187

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

M
Miklos Szeredi 已提交
1188
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1189 1190 1191 1192 1193 1194 1195
	if (err)
		goto out;

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

	err = nfserr_notdir;
A
Al Viro 已提交
1196
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1197 1198 1199 1200 1201 1202
		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) {
1203 1204 1205 1206
		host_err = fh_want_write(fhp);
		if (host_err)
			goto out_nfserr;

L
Linus Torvalds 已提交
1207
		/* called from nfsd_proc_mkdir, or possibly nfsd3_proc_create */
P
Peter Zijlstra 已提交
1208
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1209
		dchild = lookup_one_len(fname, dentry, flen);
1210
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
		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
A
Al Viro 已提交
1222 1223
				"nfsd_create: parent %pd2 not locked!\n",
				dentry);
A
Al Viro 已提交
1224
			err = nfserr_io;
L
Linus Torvalds 已提交
1225 1226 1227 1228 1229 1230 1231 1232
			goto out;
		}
	}
	/*
	 * Make sure the child dentry is still negative ...
	 */
	err = nfserr_exist;
	if (dchild->d_inode) {
A
Al Viro 已提交
1233 1234
		dprintk("nfsd_create: dentry %pd/%pd not negative!\n",
			dentry, dchild);
L
Linus Torvalds 已提交
1235 1236 1237 1238 1239 1240 1241
		goto out; 
	}

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

1242 1243 1244 1245 1246 1247 1248
	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;
	}

L
Linus Torvalds 已提交
1249 1250 1251
	/*
	 * Get the dir op function pointer.
	 */
1252
	err = 0;
1253
	host_err = 0;
L
Linus Torvalds 已提交
1254 1255
	switch (type) {
	case S_IFREG:
A
Al Viro 已提交
1256
		host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1257 1258
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
L
Linus Torvalds 已提交
1259 1260
		break;
	case S_IFDIR:
1261
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1262 1263 1264 1265 1266
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1267
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1268 1269
		break;
	}
1270
	if (host_err < 0)
L
Linus Torvalds 已提交
1271 1272
		goto out_nfserr;

1273
	err = nfsd_create_setattr(rqstp, resfhp, iap);
L
Linus Torvalds 已提交
1274

1275 1276 1277 1278 1279 1280
	/*
	 * 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));
1281 1282
	if (err2)
		err = err2;
L
Linus Torvalds 已提交
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293
	/*
	 * 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:
1294
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1295 1296 1297 1298
	goto out;
}

#ifdef CONFIG_NFSD_V3
1299 1300 1301 1302 1303 1304 1305

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

L
Linus Torvalds 已提交
1306
/*
1307
 * NFSv3 and NFSv4 version of nfsd_create
L
Linus Torvalds 已提交
1308
 */
1309
__be32
1310
do_nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
L
Linus Torvalds 已提交
1311 1312
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1313
	        bool *truncp, bool *created)
L
Linus Torvalds 已提交
1314 1315 1316
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1317 1318
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
	__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;
1329
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
1330 1331 1332 1333 1334 1335 1336 1337 1338
	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
1339
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1340
		goto out;
1341 1342 1343 1344 1345

	host_err = fh_want_write(fhp);
	if (host_err)
		goto out_nfserr;

P
Peter Zijlstra 已提交
1346
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1352
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1353 1354 1355
	if (IS_ERR(dchild))
		goto out_nfserr;

1356 1357 1358 1359 1360 1361 1362
	/* If file doesn't exist, check for permissions to create one */
	if (!dchild->d_inode) {
		err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
		if (err)
			goto out;
	}

L
Linus Torvalds 已提交
1363 1364 1365 1366
	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
	if (err)
		goto out;

1367
	if (nfsd_create_is_exclusive(createmode)) {
1368
		/* solaris7 gets confused (bugid 4218508) if these have
1369 1370 1371 1372
		 * 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
1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
	if (dchild->d_inode) {
		err = 0;

		switch (createmode) {
		case NFS3_CREATE_UNCHECKED:
			if (! S_ISREG(dchild->d_inode->i_mode))
1384
				goto out;
L
Linus Torvalds 已提交
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
			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
1403 1404 1405
			    && dchild->d_inode->i_size  == 0 ) {
				if (created)
					*created = 1;
L
Linus Torvalds 已提交
1406
				break;
1407
			}
1408 1409 1410
		case NFS4_CREATE_EXCLUSIVE4_1:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
1411 1412 1413
			    && dchild->d_inode->i_size  == 0 ) {
				if (created)
					*created = 1;
1414
				goto set_attr;
1415
			}
L
Linus Torvalds 已提交
1416 1417 1418 1419
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
A
Al Viro 已提交
1420
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1421 1422 1423
		goto out;
	}

A
Al Viro 已提交
1424
	host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1425
	if (host_err < 0) {
A
Al Viro 已提交
1426
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1427
		goto out_nfserr;
1428
	}
1429 1430
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1431

1432 1433
	nfsd_check_ignore_resizing(iap);

1434
	if (nfsd_create_is_exclusive(createmode)) {
1435
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1436
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1437
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1438 1439 1440 1441 1442 1443 1444 1445
		/* 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:
1446 1447 1448 1449 1450 1451 1452
	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));
1453 1454 1455 1456 1457 1458

	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1459 1460 1461 1462 1463

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
1464
	fh_drop_write(fhp);
L
Linus Torvalds 已提交
1465 1466 1467
 	return err;
 
 out_nfserr:
1468
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1469 1470 1471 1472 1473 1474 1475 1476 1477
	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
 */
1478
__be32
L
Linus Torvalds 已提交
1479 1480 1481 1482
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct inode	*inode;
	mm_segment_t	oldfs;
1483 1484
	__be32		err;
	int		host_err;
A
Al Viro 已提交
1485
	struct path path;
L
Linus Torvalds 已提交
1486

M
Miklos Szeredi 已提交
1487
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1488 1489 1490
	if (err)
		goto out;

A
Al Viro 已提交
1491 1492 1493
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
1494 1495

	err = nfserr_inval;
A
Al Viro 已提交
1496
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1497 1498
		goto out;

A
Al Viro 已提交
1499
	touch_atime(&path);
L
Linus Torvalds 已提交
1500 1501 1502 1503 1504
	/* 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);
1505
	host_err = inode->i_op->readlink(path.dentry, (char __user *)buf, *lenp);
L
Linus Torvalds 已提交
1506 1507
	set_fs(oldfs);

1508
	if (host_err < 0)
L
Linus Torvalds 已提交
1509
		goto out_nfserr;
1510
	*lenp = host_err;
L
Linus Torvalds 已提交
1511 1512 1513 1514 1515
	err = 0;
out:
	return err;

out_nfserr:
1516
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1517 1518 1519 1520 1521 1522 1523
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1524
__be32
L
Linus Torvalds 已提交
1525 1526 1527 1528 1529 1530 1531
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;
1532 1533
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1534 1535 1536 1537 1538 1539 1540 1541

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

M
Miklos Szeredi 已提交
1542
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1543 1544
	if (err)
		goto out;
1545 1546 1547 1548 1549

	host_err = fh_want_write(fhp);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1550 1551 1552
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1553
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1554 1555 1556 1557 1558 1559
	if (IS_ERR(dnew))
		goto out_nfserr;

	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1560
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1561 1562 1563
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1564
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1565 1566 1567
			kfree(path_alloced);
		}
	} else
1568
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
1569
	err = nfserrno(host_err);
1570 1571
	if (!err)
		err = nfserrno(commit_metadata(fhp));
L
Linus Torvalds 已提交
1572 1573
	fh_unlock(fhp);

A
Al Viro 已提交
1574
	fh_drop_write(fhp);
1575

L
Linus Torvalds 已提交
1576 1577 1578 1579 1580 1581 1582
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1583
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1584 1585 1586 1587 1588 1589 1590
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1591
__be32
L
Linus Torvalds 已提交
1592 1593 1594 1595
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
1596
	struct inode	*dirp;
1597 1598
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1599

M
Miklos Szeredi 已提交
1600
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1601 1602
	if (err)
		goto out;
1603
	err = fh_verify(rqstp, tfhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1604 1605
	if (err)
		goto out;
1606 1607 1608
	err = nfserr_isdir;
	if (S_ISDIR(tfhp->fh_dentry->d_inode->i_mode))
		goto out;
L
Linus Torvalds 已提交
1609 1610 1611 1612 1613 1614 1615
	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

1616 1617 1618 1619 1620 1621
	host_err = fh_want_write(tfhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

P
Peter Zijlstra 已提交
1622
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1623 1624 1625 1626
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1627
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1628 1629 1630 1631 1632
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;

1633 1634
	err = nfserr_noent;
	if (!dold->d_inode)
1635
		goto out_dput;
J
J. Bruce Fields 已提交
1636
	host_err = vfs_link(dold, dirp, dnew, NULL);
1637
	if (!host_err) {
1638 1639 1640
		err = nfserrno(commit_metadata(ffhp));
		if (!err)
			err = nfserrno(commit_metadata(tfhp));
L
Linus Torvalds 已提交
1641
	} else {
1642
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1643 1644
			err = nfserr_acces;
		else
1645
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1646
	}
1647
out_dput:
L
Linus Torvalds 已提交
1648
	dput(dnew);
1649 1650
out_unlock:
	fh_unlock(ffhp);
1651
	fh_drop_write(tfhp);
L
Linus Torvalds 已提交
1652 1653 1654 1655
out:
	return err;

out_nfserr:
1656
	err = nfserrno(host_err);
1657
	goto out_unlock;
L
Linus Torvalds 已提交
1658 1659 1660 1661 1662 1663
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1664
__be32
L
Linus Torvalds 已提交
1665 1666 1667 1668 1669
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;
1670 1671
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1672

M
Miklos Szeredi 已提交
1673
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1674 1675
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1676
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689
	if (err)
		goto out;

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

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

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

1690 1691 1692 1693 1694 1695
	host_err = fh_want_write(ffhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

L
Linus Torvalds 已提交
1696 1697 1698 1699 1700 1701 1702 1703
	/* 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);
1704
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1705 1706 1707
	if (IS_ERR(odentry))
		goto out_nfserr;

1708
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1709 1710
	if (!odentry->d_inode)
		goto out_dput_old;
1711
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1712 1713 1714 1715
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1716
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1717 1718
	if (IS_ERR(ndentry))
		goto out_dput_old;
1719
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1720 1721 1722
	if (ndentry == trap)
		goto out_dput_new;

1723 1724 1725
	host_err = -EXDEV;
	if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
		goto out_dput_new;
1726 1727
	if (ffhp->fh_export->ex_path.dentry != tfhp->fh_export->ex_path.dentry)
		goto out_dput_new;
1728

M
Miklos Szeredi 已提交
1729
	host_err = vfs_rename(fdir, odentry, tdir, ndentry, NULL, 0);
1730 1731
	if (!host_err) {
		host_err = commit_metadata(tfhp);
1732
		if (!host_err)
1733
			host_err = commit_metadata(ffhp);
L
Linus Torvalds 已提交
1734 1735 1736 1737 1738 1739
	}
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1740
	err = nfserrno(host_err);
1741 1742
	/*
	 * We cannot rely on fh_unlock on the two filehandles,
L
Linus Torvalds 已提交
1743
	 * as that would do the wrong thing if the two directories
1744
	 * were the same, so again we do it by hand.
L
Linus Torvalds 已提交
1745 1746 1747 1748 1749
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;
1750
	fh_drop_write(ffhp);
L
Linus Torvalds 已提交
1751 1752 1753 1754 1755 1756 1757 1758 1759

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
1760
__be32
L
Linus Torvalds 已提交
1761 1762 1763 1764 1765
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1766 1767
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1768 1769 1770 1771

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1772
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1773 1774 1775
	if (err)
		goto out;

1776 1777 1778 1779
	host_err = fh_want_write(fhp);
	if (host_err)
		goto out_nfserr;

P
Peter Zijlstra 已提交
1780
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1781 1782 1783 1784
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1785
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797
	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;

1798
	if (type != S_IFDIR)
1799
		host_err = vfs_unlink(dirp, rdentry, NULL);
1800
	else
1801
		host_err = vfs_rmdir(dirp, rdentry);
1802 1803
	if (!host_err)
		host_err = commit_metadata(fhp);
1804 1805
	dput(rdentry);

L
Linus Torvalds 已提交
1806
out_nfserr:
1807
	err = nfserrno(host_err);
1808 1809
out:
	return err;
L
Linus Torvalds 已提交
1810 1811
}

1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827
/*
 * 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 {
1828
	struct dir_context ctx;
1829 1830
	char		*dirent;
	size_t		used;
1831
	int		full;
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841
};

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));
1842 1843
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1844
		return -EINVAL;
1845
	}
1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856

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

	return 0;
}

1857 1858
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1859
{
1860
	struct buffered_dirent *de;
1861
	int host_err;
1862 1863
	int size;
	loff_t offset;
A
Al Viro 已提交
1864 1865 1866 1867
	struct readdir_data buf = {
		.ctx.actor = nfsd_buffered_filldir,
		.dirent = (void *)__get_free_page(GFP_KERNEL)
	};
1868

1869
	if (!buf.dirent)
1870
		return nfserrno(-ENOMEM);
1871 1872

	offset = *offsetp;
1873

1874
	while (1) {
A
Al Viro 已提交
1875
		struct inode *dir_inode = file_inode(file);
1876 1877
		unsigned int reclen;

1878
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1879
		buf.used = 0;
1880
		buf.full = 0;
1881

1882
		host_err = iterate_dir(file, &buf.ctx);
1883 1884 1885 1886
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1887 1888 1889 1890 1891 1892 1893
			break;

		size = buf.used;

		if (!size)
			break;

1894 1895 1896 1897 1898 1899 1900 1901 1902
		/*
		 * 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;

1903 1904 1905 1906 1907 1908
		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))
1909
				break;
1910 1911

			if (cdp->err != nfs_ok)
1912
				break;
1913 1914 1915 1916 1917 1918

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1919 1920 1921 1922
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1923
		offset = vfs_llseek(file, 0, SEEK_CUR);
1924 1925 1926
	}

	free_page((unsigned long)(buf.dirent));
1927 1928 1929

	if (host_err)
		return nfserrno(host_err);
1930 1931 1932

	*offsetp = offset;
	return cdp->err;
1933 1934
}

L
Linus Torvalds 已提交
1935 1936 1937 1938
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1939
__be32
L
Linus Torvalds 已提交
1940
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1941
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1942
{
1943
	__be32		err;
L
Linus Torvalds 已提交
1944 1945
	struct file	*file;
	loff_t		offset = *offsetp;
1946 1947 1948 1949 1950
	int             may_flags = NFSD_MAY_READ;

	/* NFSv2 only supports 32 bit cookies */
	if (rqstp->rq_vers > 2)
		may_flags |= NFSD_MAY_64BIT_COOKIE;
L
Linus Torvalds 已提交
1951

1952
	err = nfsd_open(rqstp, fhp, S_IFDIR, may_flags, &file);
L
Linus Torvalds 已提交
1953 1954 1955
	if (err)
		goto out;

1956
	offset = vfs_llseek(file, offset, SEEK_SET);
L
Linus Torvalds 已提交
1957 1958 1959 1960 1961
	if (offset < 0) {
		err = nfserrno((int)offset);
		goto out_close;
	}

1962
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975

	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
 */
1976
__be32
1977
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
1978
{
1979 1980 1981
	__be32 err;

	err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
1982 1983 1984 1985 1986 1987 1988 1989
	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
1990 1991 1992
	return err;
}

J
J. Bruce Fields 已提交
1993
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
1994
{
J
J. Bruce Fields 已提交
1995
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
1996 1997
}

L
Linus Torvalds 已提交
1998 1999 2000
/*
 * Check for a user's access permissions to this inode.
 */
2001
__be32
2002 2003
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
2004 2005 2006 2007
{
	struct inode	*inode = dentry->d_inode;
	int		err;

2008
	if ((acc & NFSD_MAY_MASK) == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
2009 2010 2011 2012
		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
2013 2014 2015 2016 2017 2018 2019
		(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
2020 2021 2022
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2023
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2024
	dprintk("      owner %d/%d user %d/%d\n",
2025
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2026 2027 2028 2029 2030 2031
#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
2032 2033
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2034 2035
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2036
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2037
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2038 2039
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2040
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2041 2042
		return nfserr_perm;

M
Miklos Szeredi 已提交
2043
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2044 2045 2046 2047 2048 2049 2050
		/* 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
2051
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066
	}
	/*
	 * 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
2067
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2068
	    uid_eq(inode->i_uid, current_fsuid()))
L
Linus Torvalds 已提交
2069 2070
		return 0;

M
Miklos Szeredi 已提交
2071
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2072
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2073 2074 2075

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
2076 2077
	     (acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE) ||
	      acc == (NFSD_MAY_READ | NFSD_MAY_READ_IF_EXEC)))
2078
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
2079 2080 2081 2082 2083 2084 2085

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2086 2087 2088
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2089
	dprintk("nfsd: freeing readahead buffers.\n");
2090 2091 2092 2093 2094 2095 2096 2097 2098 2099

	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
2100 2101 2102 2103 2104 2105 2106 2107
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2108 2109
	int	j = 0;
	int	nperbucket;
2110
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2111

2112

2113
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2114
		return 0;
2115 2116 2117 2118
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2119 2120

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2121 2122

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2123
		spin_lock_init(&raparm_hash[i].pb_lock);
2124 2125 2126 2127 2128 2129 2130 2131 2132

		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;
2133 2134
	}

L
Linus Torvalds 已提交
2135 2136
	nfsdstats.ra_size = cache_size;
	return 0;
2137 2138 2139 2140 2141

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2142
}