vfs.c 50.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * 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

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

	dparent = fhp->fh_dentry;
192
	exp = exp_get(fhp->fh_export);
L
Linus Torvalds 已提交
193 194 195 196 197

	/* Lookup the name, but don't follow links */
	if (isdotent(name, len)) {
		if (len==1)
			dentry = dget(dparent);
198
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
199
			dentry = dget_parent(dparent);
200
		else if (!EX_NOHIDE(exp) && !nfsd_v4client(rqstp))
L
Linus Torvalds 已提交
201 202 203
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
204 205
			host_err = nfsd_lookup_parent(rqstp, dparent, &exp, &dentry);
			if (host_err)
L
Linus Torvalds 已提交
206 207 208
				goto out_nfserr;
		}
	} else {
209 210 211 212 213 214
		/*
		 * 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
215
		dentry = lookup_one_len(name, dparent, len);
216
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
217 218 219 220 221
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
222
		if (nfsd_mountpoint(dentry, exp)) {
223
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
224 225 226 227 228
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251
	*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,
252
				unsigned int len, struct svc_fh *resfh)
253 254 255 256 257
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

258 259 260
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
	if (err)
		return err;
261 262 263
	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
264 265 266
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
267 268 269 270 271 272 273
	/*
	 * 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;
274
out:
L
Linus Torvalds 已提交
275 276 277 278 279
	dput(dentry);
	exp_put(exp);
	return err;
}

280 281 282 283 284 285 286 287 288 289 290 291
/*
 * 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
292 293 294
	if (export_ops->commit_metadata)
		return export_ops->commit_metadata(inode);
	return sync_inode_metadata(inode, 1);
295
}
296

L
Linus Torvalds 已提交
297
/*
C
Christoph Hellwig 已提交
298 299
 * Go over the attributes and take care of the small differences between
 * NFS semantics and what Linux expects.
L
Linus Torvalds 已提交
300
 */
C
Christoph Hellwig 已提交
301 302
static void
nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
L
Linus Torvalds 已提交
303
{
304 305
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
306 307 308 309 310 311 312
	 * 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
313
	 * it is not an interface that we should be using.
L
Linus Torvalds 已提交
314 315 316
	 */
#define BOTH_TIME_SET (ATTR_ATIME_SET | ATTR_MTIME_SET)
#define	MAX_TOUCH_TIME_ERROR (30*60)
317 318 319 320 321 322 323 324
	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
325
		 */
326 327 328 329 330 331 332 333 334 335 336 337
		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
338 339
	}

340
	/* sanitize the mode change */
L
Linus Torvalds 已提交
341 342
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
343
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
344 345 346
	}

	/* Revoke setuid/setgid on chown */
347
	if (!S_ISDIR(inode->i_mode) &&
348
	    ((iap->ia_valid & ATTR_UID) || (iap->ia_valid & ATTR_GID))) {
349 350 351
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
352
			iap->ia_mode &= ~S_ISUID;
353 354 355 356 357
			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);
358
		}
L
Linus Torvalds 已提交
359
	}
C
Christoph Hellwig 已提交
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
}

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;
406
	bool		get_write_count;
C
Christoph Hellwig 已提交
407 408 409 410 411 412 413
	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;

414 415 416
	/* Callers that do fh_verify should do the fh_want_write: */
	get_write_count = !fhp->fh_dentry;

C
Christoph Hellwig 已提交
417 418 419 420
	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
	if (err)
		goto out;
421 422 423 424 425
	if (get_write_count) {
		host_err = fh_want_write(fhp);
		if (host_err)
			return nfserrno(host_err);
	}
C
Christoph Hellwig 已提交
426 427 428 429 430 431 432 433 434 435

	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
436

C
Christoph Hellwig 已提交
437 438 439 440 441 442 443 444 445 446 447 448
	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
449 450 451

	iap->ia_valid |= ATTR_CTIME;

452 453 454
	if (check_guard && guardtime != inode->i_ctime.tv_sec) {
		err = nfserr_notsync;
		goto out_put_write_access;
L
Linus Torvalds 已提交
455
	}
456 457 458 459

	fh_lock(fhp);
	host_err = notify_change(dentry, iap, NULL);
	fh_unlock(fhp);
460
	err = nfserrno(host_err);
461 462

out_put_write_access:
L
Linus Torvalds 已提交
463 464 465
	if (size_change)
		put_write_access(inode);
	if (!err)
466
		commit_metadata(fhp);
L
Linus Torvalds 已提交
467 468 469 470
out:
	return err;
}

471
#if defined(CONFIG_NFSD_V4)
472 473 474 475 476 477 478 479 480 481 482 483 484
/*
 * 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.
 */
485 486 487 488 489 490 491 492 493 494
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;
495
	if (vfs_getxattr(dentry, NFSD_JUNCTION_XATTR_NAME, NULL, 0) <= 0)
496 497 498
		return 0;
	return 1;
}
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525
#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
526
#endif /* defined(CONFIG_NFSD_V4) */
L
Linus Torvalds 已提交
527 528 529 530 531 532 533 534 535 536

#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
537 538 539 540
    {	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
541 542 543 544 545

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
546 547 548 549 550
    {	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
551 552 553 554 555 556 557 558 559 560 561 562

    {	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
563 564 565 566
    {	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
567 568 569 570

    {	0,			0				}
};

571
__be32
L
Linus Torvalds 已提交
572 573 574 575 576 577
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;
578
	__be32			error;
L
Linus Torvalds 已提交
579

M
Miklos Szeredi 已提交
580
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
	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) {
598
			__be32 err2;
L
Linus Torvalds 已提交
599 600 601

			sresult |= map->access;

602
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
			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 */

630 631 632
static int nfsd_open_break_lease(struct inode *inode, int access)
{
	unsigned int mode;
L
Linus Torvalds 已提交
633

634 635 636 637 638
	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
639 640 641

/*
 * Open an existing file or directory.
642 643
 * The may_flags argument indicates the type of open (read/write/lock)
 * and additional flags.
L
Linus Torvalds 已提交
644 645
 * N.B. After this call fhp needs an fh_put
 */
646
__be32
A
Al Viro 已提交
647
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
648
			int may_flags, struct file **filp)
L
Linus Torvalds 已提交
649
{
650
	struct path	path;
L
Linus Torvalds 已提交
651
	struct inode	*inode;
652 653
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
654
	int		host_err = 0;
L
Linus Torvalds 已提交
655

656 657
	validate_process_creds();

L
Linus Torvalds 已提交
658 659 660 661
	/*
	 * 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.
662 663 664 665 666 667
	 *
	 * 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
668
	 */
669 670 671
	if (type == S_IFREG)
		may_flags |= NFSD_MAY_OWNER_OVERRIDE;
	err = fh_verify(rqstp, fhp, type, may_flags);
L
Linus Torvalds 已提交
672 673 674
	if (err)
		goto out;

675 676 677
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
678 679 680 681 682

	/* Disallow write access to files with the append-only bit set
	 * or any access when mandatory locking enabled
	 */
	err = nfserr_perm;
683
	if (IS_APPEND(inode) && (may_flags & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
684
		goto out;
J
J. Bruce Fields 已提交
685 686 687 688 689 690
	/*
	 * 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
691 692 693 694 695
		goto out;

	if (!inode->i_fop)
		goto out;

696
	host_err = nfsd_open_break_lease(inode, may_flags);
697
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
698 699
		goto out_nfserr;

700 701
	if (may_flags & NFSD_MAY_WRITE) {
		if (may_flags & NFSD_MAY_READ)
702 703 704
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
705
	}
706
	*filp = dentry_open(&path, flags, current_cred());
707
	if (IS_ERR(*filp)) {
708
		host_err = PTR_ERR(*filp);
709 710
		*filp = NULL;
	} else {
711 712
		host_err = ima_file_check(*filp, may_flags);

713 714 715 716 717 718
		if (may_flags & NFSD_MAY_64BIT_COOKIE)
			(*filp)->f_mode |= FMODE_64BITHASH;
		else
			(*filp)->f_mode |= FMODE_32BITHASH;
	}

L
Linus Torvalds 已提交
719
out_nfserr:
720
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
721
out:
722
	validate_process_creds();
L
Linus Torvalds 已提交
723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
	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;
745 746 747 748 749
	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
750

751 752
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
753 754 755 756 757 758
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
759
	depth = nfsdstats.ra_size;
L
Linus Torvalds 已提交
760
	if (!frap) {	
761
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
762 763 764 765 766 767 768
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
769
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
770
found:
771
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
772
		*rap = ra->p_next;
773 774
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
775 776 777
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
778
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
779 780 781 782
	return ra;
}

/*
783 784 785
 * 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
786 787
 */
static int
788 789
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
790
{
791
	struct svc_rqst *rqstp = sd->u.data;
792
	struct page **pp = rqstp->rq_next_page;
793 794 795 796
	struct page *page = buf->page;
	size_t size;

	size = sd->len;
L
Linus Torvalds 已提交
797 798 799

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
800 801
		put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
802
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
803
		rqstp->rq_res.page_len = size;
804
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
805
		get_page(page);
806 807 808
		if (*rqstp->rq_next_page)
			put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
L
Linus Torvalds 已提交
809
		rqstp->rq_res.page_len += size;
810
	} else
L
Linus Torvalds 已提交
811 812 813 814 815
		rqstp->rq_res.page_len += size;

	return size;
}

816 817 818 819 820 821
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

822
__be32 nfsd_finish_read(struct file *file, unsigned long *count, int host_err)
L
Linus Torvalds 已提交
823
{
824 825 826
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
827
		fsnotify_access(file);
828
		return 0;
L
Linus Torvalds 已提交
829
	} else 
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869
		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
870 871
}

872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887
/*
 * 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
888
	struct inode *inode = file_inode(file);
889 890 891 892 893 894 895 896 897 898 899 900 901
	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));
902
		err = vfs_fsync(file, 0);
903 904 905 906 907 908
	}
	last_ino = inode->i_ino;
	last_dev = inode->i_sb->s_dev;
	return err;
}

909
static __be32
L
Linus Torvalds 已提交
910 911
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
912
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
913 914 915 916 917
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
918 919
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
920
	int			stable = *stablep;
921
	int			use_wgather;
K
Kent Overstreet 已提交
922
	loff_t			pos = offset;
923 924 925 926 927 928 929 930 931 932
	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
933

934
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
935 936 937
	inode = dentry->d_inode;
	exp   = fhp->fh_export;

938
	use_wgather = (rqstp->rq_vers == 2) && EX_WGATHER(exp);
L
Linus Torvalds 已提交
939 940 941 942 943 944

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

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
K
Kent Overstreet 已提交
945
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &pos);
L
Linus Torvalds 已提交
946
	set_fs(oldfs);
947 948 949 950
	if (host_err < 0)
		goto out_nfserr;
	*cnt = host_err;
	nfsdstats.io_write += host_err;
951
	fsnotify_modify(file);
L
Linus Torvalds 已提交
952

953 954 955 956 957 958
	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
959

960
out_nfserr:
961
	dprintk("nfsd: write complete host_err=%d\n", host_err);
962
	if (host_err >= 0)
L
Linus Torvalds 已提交
963
		err = 0;
964
	else
965
		err = nfserrno(host_err);
966 967
	if (rqstp->rq_local)
		tsk_restore_flags(current, pflags, PF_LESS_THROTTLE);
L
Linus Torvalds 已提交
968 969 970
	return err;
}

971 972
__be32 nfsd_get_tmp_read_open(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct file **file, struct raparms **ra)
973 974 975 976
{
	struct inode *inode;
	__be32 err;

977
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, file);
978 979 980
	if (err)
		return err;

981
	inode = file_inode(*file);
982 983

	/* Get readahead parameters */
984
	*ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);
985

986 987 988 989
	if (*ra && (*ra)->p_set)
		(*file)->f_ra = (*ra)->p_ra;
	return nfs_ok;
}
990

991 992
void nfsd_put_tmp_read_open(struct file *file, struct raparms *ra)
{
993 994 995 996 997 998 999 1000 1001 1002
	/* 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);
1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
}

/*
 * 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);

1025 1026 1027
	return err;
}

L
Linus Torvalds 已提交
1028 1029 1030 1031 1032
/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1033
__be32
L
Linus Torvalds 已提交
1034
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1035
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1036 1037
		int *stablep)
{
1038
	__be32			err = 0;
L
Linus Torvalds 已提交
1039 1040

	if (file) {
1041
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1042
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1043 1044 1045 1046 1047
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1048
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
		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.
1064 1065 1066
 *
 * 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
1067 1068 1069 1070
 *
 * 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.
 */
1071
__be32
L
Linus Torvalds 已提交
1072 1073 1074 1075
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1076 1077
	loff_t		end = LLONG_MAX;
	__be32		err = nfserr_inval;
L
Linus Torvalds 已提交
1078

1079 1080 1081 1082 1083 1084 1085
	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
1086

1087 1088
	err = nfsd_open(rqstp, fhp, S_IFREG,
			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &file);
M
Miklos Szeredi 已提交
1089
	if (err)
1090
		goto out;
L
Linus Torvalds 已提交
1091
	if (EX_ISSYNC(fhp->fh_export)) {
1092
		int err2 = vfs_fsync_range(file, offset, end, 0);
1093 1094 1095 1096

		if (err2 != -EINVAL)
			err = nfserrno(err2);
		else
L
Linus Torvalds 已提交
1097 1098 1099 1100
			err = nfserr_notsupp;
	}

	nfsd_close(file);
1101
out:
L
Linus Torvalds 已提交
1102 1103 1104 1105
	return err;
}
#endif /* CONFIG_NFSD_V3 */

A
Adrian Bunk 已提交
1106
static __be32
1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
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:
	 */
1119
	if (!uid_eq(current_fsuid(), GLOBAL_ROOT_UID))
1120 1121 1122 1123 1124 1125
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140
/* 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
1141 1142 1143 1144 1145 1146 1147 1148
/*
 * 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
 */
1149
__be32
L
Linus Torvalds 已提交
1150 1151 1152 1153 1154 1155
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;
1156
	__be32		err;
1157
	__be32		err2;
1158
	int		host_err;
L
Linus Torvalds 已提交
1159 1160 1161 1162 1163 1164 1165 1166

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

M
Miklos Szeredi 已提交
1167
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1168 1169 1170 1171 1172 1173 1174
	if (err)
		goto out;

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

	err = nfserr_notdir;
A
Al Viro 已提交
1175
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181
		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) {
1182 1183 1184 1185
		host_err = fh_want_write(fhp);
		if (host_err)
			goto out_nfserr;

L
Linus Torvalds 已提交
1186
		/* called from nfsd_proc_mkdir, or possibly nfsd3_proc_create */
P
Peter Zijlstra 已提交
1187
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1188
		dchild = lookup_one_len(fname, dentry, flen);
1189
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200
		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
1201 1202
				"nfsd_create: parent %pd2 not locked!\n",
				dentry);
A
Al Viro 已提交
1203
			err = nfserr_io;
L
Linus Torvalds 已提交
1204 1205 1206 1207 1208 1209 1210 1211
			goto out;
		}
	}
	/*
	 * Make sure the child dentry is still negative ...
	 */
	err = nfserr_exist;
	if (dchild->d_inode) {
A
Al Viro 已提交
1212 1213
		dprintk("nfsd_create: dentry %pd/%pd not negative!\n",
			dentry, dchild);
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219 1220
		goto out; 
	}

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

1221 1222 1223 1224 1225 1226 1227
	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
1228 1229 1230
	/*
	 * Get the dir op function pointer.
	 */
1231
	err = 0;
1232
	host_err = 0;
L
Linus Torvalds 已提交
1233 1234
	switch (type) {
	case S_IFREG:
A
Al Viro 已提交
1235
		host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1236 1237
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
L
Linus Torvalds 已提交
1238 1239
		break;
	case S_IFDIR:
1240
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1246
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1247 1248
		break;
	}
1249
	if (host_err < 0)
L
Linus Torvalds 已提交
1250 1251
		goto out_nfserr;

1252
	err = nfsd_create_setattr(rqstp, resfhp, iap);
L
Linus Torvalds 已提交
1253

1254 1255 1256 1257 1258 1259
	/*
	 * 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));
1260 1261
	if (err2)
		err = err2;
L
Linus Torvalds 已提交
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
	/*
	 * 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:
1273
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1274 1275 1276 1277
	goto out;
}

#ifdef CONFIG_NFSD_V3
1278 1279 1280 1281 1282 1283 1284

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
1285
/*
1286
 * NFSv3 and NFSv4 version of nfsd_create
L
Linus Torvalds 已提交
1287
 */
1288
__be32
1289
do_nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
L
Linus Torvalds 已提交
1290 1291
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1292
	        bool *truncp, bool *created)
L
Linus Torvalds 已提交
1293 1294 1295
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1296 1297
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307
	__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;
1308
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
1309 1310 1311 1312 1313 1314 1315 1316 1317
	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
1318
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1319
		goto out;
1320 1321 1322 1323 1324

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

P
Peter Zijlstra 已提交
1325
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1326 1327 1328 1329 1330

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1331
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1332 1333 1334
	if (IS_ERR(dchild))
		goto out_nfserr;

1335 1336 1337 1338 1339 1340 1341
	/* 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
1342 1343 1344 1345
	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
	if (err)
		goto out;

1346
	if (nfsd_create_is_exclusive(createmode)) {
1347
		/* solaris7 gets confused (bugid 4218508) if these have
1348 1349 1350 1351
		 * 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
1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
		 */
		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))
1363
				goto out;
L
Linus Torvalds 已提交
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
			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
1382 1383 1384
			    && dchild->d_inode->i_size  == 0 ) {
				if (created)
					*created = 1;
L
Linus Torvalds 已提交
1385
				break;
1386
			}
1387 1388 1389
		case NFS4_CREATE_EXCLUSIVE4_1:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
1390 1391 1392
			    && dchild->d_inode->i_size  == 0 ) {
				if (created)
					*created = 1;
1393
				goto set_attr;
1394
			}
L
Linus Torvalds 已提交
1395 1396 1397 1398
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
A
Al Viro 已提交
1399
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1400 1401 1402
		goto out;
	}

A
Al Viro 已提交
1403
	host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1404
	if (host_err < 0) {
A
Al Viro 已提交
1405
		fh_drop_write(fhp);
L
Linus Torvalds 已提交
1406
		goto out_nfserr;
1407
	}
1408 1409
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1410

1411 1412
	nfsd_check_ignore_resizing(iap);

1413
	if (nfsd_create_is_exclusive(createmode)) {
1414
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1415
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1416
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1417 1418 1419 1420 1421 1422 1423 1424
		/* 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:
1425 1426 1427 1428 1429 1430 1431
	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));
1432 1433 1434 1435 1436 1437

	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1438 1439 1440 1441 1442

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
1443
	fh_drop_write(fhp);
L
Linus Torvalds 已提交
1444 1445 1446
 	return err;
 
 out_nfserr:
1447
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1448 1449 1450 1451 1452 1453 1454 1455 1456
	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
 */
1457
__be32
L
Linus Torvalds 已提交
1458 1459 1460 1461
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct inode	*inode;
	mm_segment_t	oldfs;
1462 1463
	__be32		err;
	int		host_err;
A
Al Viro 已提交
1464
	struct path path;
L
Linus Torvalds 已提交
1465

M
Miklos Szeredi 已提交
1466
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1467 1468 1469
	if (err)
		goto out;

A
Al Viro 已提交
1470 1471 1472
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
1473 1474

	err = nfserr_inval;
A
Al Viro 已提交
1475
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1476 1477
		goto out;

A
Al Viro 已提交
1478
	touch_atime(&path);
L
Linus Torvalds 已提交
1479 1480 1481 1482 1483
	/* 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);
1484
	host_err = inode->i_op->readlink(path.dentry, (char __user *)buf, *lenp);
L
Linus Torvalds 已提交
1485 1486
	set_fs(oldfs);

1487
	if (host_err < 0)
L
Linus Torvalds 已提交
1488
		goto out_nfserr;
1489
	*lenp = host_err;
L
Linus Torvalds 已提交
1490 1491 1492 1493 1494
	err = 0;
out:
	return err;

out_nfserr:
1495
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1496 1497 1498 1499 1500 1501 1502
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1503
__be32
L
Linus Torvalds 已提交
1504 1505 1506 1507 1508 1509 1510
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;
1511 1512
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1513 1514 1515 1516 1517 1518 1519 1520

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

M
Miklos Szeredi 已提交
1521
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1522 1523
	if (err)
		goto out;
1524 1525 1526 1527 1528

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

L
Linus Torvalds 已提交
1529 1530 1531
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1532
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1533 1534 1535 1536 1537 1538
	if (IS_ERR(dnew))
		goto out_nfserr;

	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1539
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1540 1541 1542
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1543
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1544 1545 1546
			kfree(path_alloced);
		}
	} else
1547
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
1548
	err = nfserrno(host_err);
1549 1550
	if (!err)
		err = nfserrno(commit_metadata(fhp));
L
Linus Torvalds 已提交
1551 1552
	fh_unlock(fhp);

A
Al Viro 已提交
1553
	fh_drop_write(fhp);
1554

L
Linus Torvalds 已提交
1555 1556 1557 1558 1559 1560 1561
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1562
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1563 1564 1565 1566 1567 1568 1569
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1570
__be32
L
Linus Torvalds 已提交
1571 1572 1573 1574
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
1575
	struct inode	*dirp;
1576 1577
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1578

M
Miklos Szeredi 已提交
1579
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1580 1581
	if (err)
		goto out;
1582
	err = fh_verify(rqstp, tfhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1583 1584
	if (err)
		goto out;
1585 1586 1587
	err = nfserr_isdir;
	if (S_ISDIR(tfhp->fh_dentry->d_inode->i_mode))
		goto out;
L
Linus Torvalds 已提交
1588 1589 1590 1591 1592 1593 1594
	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

1595 1596 1597 1598 1599 1600
	host_err = fh_want_write(tfhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

P
Peter Zijlstra 已提交
1601
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1602 1603 1604 1605
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1606
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1607 1608 1609 1610 1611
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;

1612 1613
	err = nfserr_noent;
	if (!dold->d_inode)
1614
		goto out_dput;
J
J. Bruce Fields 已提交
1615
	host_err = vfs_link(dold, dirp, dnew, NULL);
1616
	if (!host_err) {
1617 1618 1619
		err = nfserrno(commit_metadata(ffhp));
		if (!err)
			err = nfserrno(commit_metadata(tfhp));
L
Linus Torvalds 已提交
1620
	} else {
1621
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1622 1623
			err = nfserr_acces;
		else
1624
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1625
	}
1626
out_dput:
L
Linus Torvalds 已提交
1627
	dput(dnew);
1628 1629
out_unlock:
	fh_unlock(ffhp);
1630
	fh_drop_write(tfhp);
L
Linus Torvalds 已提交
1631 1632 1633 1634
out:
	return err;

out_nfserr:
1635
	err = nfserrno(host_err);
1636
	goto out_unlock;
L
Linus Torvalds 已提交
1637 1638 1639 1640 1641 1642
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1643
__be32
L
Linus Torvalds 已提交
1644 1645 1646 1647 1648
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;
1649 1650
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1651

M
Miklos Szeredi 已提交
1652
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1653 1654
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1655
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
	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;

1669 1670 1671 1672 1673 1674
	host_err = fh_want_write(ffhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

L
Linus Torvalds 已提交
1675 1676 1677 1678 1679 1680 1681 1682
	/* 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);
1683
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1684 1685 1686
	if (IS_ERR(odentry))
		goto out_nfserr;

1687
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1688 1689
	if (!odentry->d_inode)
		goto out_dput_old;
1690
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1691 1692 1693 1694
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1695
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1696 1697
	if (IS_ERR(ndentry))
		goto out_dput_old;
1698
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1699 1700 1701
	if (ndentry == trap)
		goto out_dput_new;

1702 1703 1704
	host_err = -EXDEV;
	if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
		goto out_dput_new;
1705 1706
	if (ffhp->fh_export->ex_path.dentry != tfhp->fh_export->ex_path.dentry)
		goto out_dput_new;
1707

M
Miklos Szeredi 已提交
1708
	host_err = vfs_rename(fdir, odentry, tdir, ndentry, NULL, 0);
1709 1710
	if (!host_err) {
		host_err = commit_metadata(tfhp);
1711
		if (!host_err)
1712
			host_err = commit_metadata(ffhp);
L
Linus Torvalds 已提交
1713 1714 1715 1716 1717 1718
	}
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1719
	err = nfserrno(host_err);
1720 1721
	/*
	 * We cannot rely on fh_unlock on the two filehandles,
L
Linus Torvalds 已提交
1722
	 * as that would do the wrong thing if the two directories
1723
	 * were the same, so again we do it by hand.
L
Linus Torvalds 已提交
1724 1725 1726 1727 1728
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;
1729
	fh_drop_write(ffhp);
L
Linus Torvalds 已提交
1730 1731 1732 1733 1734 1735 1736 1737 1738

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
1739
__be32
L
Linus Torvalds 已提交
1740 1741 1742 1743 1744
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1745 1746
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1747 1748 1749 1750

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1751
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1752 1753 1754
	if (err)
		goto out;

1755 1756 1757 1758
	host_err = fh_want_write(fhp);
	if (host_err)
		goto out_nfserr;

P
Peter Zijlstra 已提交
1759
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1760 1761 1762 1763
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1764
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
	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;

1777
	if (type != S_IFDIR)
1778
		host_err = vfs_unlink(dirp, rdentry, NULL);
1779
	else
1780
		host_err = vfs_rmdir(dirp, rdentry);
1781 1782
	if (!host_err)
		host_err = commit_metadata(fhp);
1783 1784
	dput(rdentry);

L
Linus Torvalds 已提交
1785
out_nfserr:
1786
	err = nfserrno(host_err);
1787 1788
out:
	return err;
L
Linus Torvalds 已提交
1789 1790
}

1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806
/*
 * 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 {
1807
	struct dir_context ctx;
1808 1809
	char		*dirent;
	size_t		used;
1810
	int		full;
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820
};

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));
1821 1822
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1823
		return -EINVAL;
1824
	}
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835

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

	return 0;
}

1836 1837
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1838
{
1839
	struct buffered_dirent *de;
1840
	int host_err;
1841 1842
	int size;
	loff_t offset;
A
Al Viro 已提交
1843 1844 1845 1846
	struct readdir_data buf = {
		.ctx.actor = nfsd_buffered_filldir,
		.dirent = (void *)__get_free_page(GFP_KERNEL)
	};
1847

1848
	if (!buf.dirent)
1849
		return nfserrno(-ENOMEM);
1850 1851

	offset = *offsetp;
1852

1853
	while (1) {
A
Al Viro 已提交
1854
		struct inode *dir_inode = file_inode(file);
1855 1856
		unsigned int reclen;

1857
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1858
		buf.used = 0;
1859
		buf.full = 0;
1860

1861
		host_err = iterate_dir(file, &buf.ctx);
1862 1863 1864 1865
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1866 1867 1868 1869 1870 1871 1872
			break;

		size = buf.used;

		if (!size)
			break;

1873 1874 1875 1876 1877 1878 1879 1880 1881
		/*
		 * 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;

1882 1883 1884 1885 1886 1887
		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))
1888
				break;
1889 1890

			if (cdp->err != nfs_ok)
1891
				break;
1892 1893 1894 1895 1896 1897

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1898 1899 1900 1901
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1902
		offset = vfs_llseek(file, 0, SEEK_CUR);
1903 1904 1905
	}

	free_page((unsigned long)(buf.dirent));
1906 1907 1908

	if (host_err)
		return nfserrno(host_err);
1909 1910 1911

	*offsetp = offset;
	return cdp->err;
1912 1913
}

L
Linus Torvalds 已提交
1914 1915 1916 1917
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1918
__be32
L
Linus Torvalds 已提交
1919
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1920
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1921
{
1922
	__be32		err;
L
Linus Torvalds 已提交
1923 1924
	struct file	*file;
	loff_t		offset = *offsetp;
1925 1926 1927 1928 1929
	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
1930

1931
	err = nfsd_open(rqstp, fhp, S_IFDIR, may_flags, &file);
L
Linus Torvalds 已提交
1932 1933 1934
	if (err)
		goto out;

1935
	offset = vfs_llseek(file, offset, SEEK_SET);
L
Linus Torvalds 已提交
1936 1937 1938 1939 1940
	if (offset < 0) {
		err = nfserrno((int)offset);
		goto out_close;
	}

1941
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954

	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
 */
1955
__be32
1956
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
1957
{
1958 1959 1960
	__be32 err;

	err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
1961 1962 1963 1964 1965 1966 1967 1968
	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
1969 1970 1971
	return err;
}

J
J. Bruce Fields 已提交
1972
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
1973
{
J
J. Bruce Fields 已提交
1974
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
1975 1976
}

L
Linus Torvalds 已提交
1977 1978 1979
/*
 * Check for a user's access permissions to this inode.
 */
1980
__be32
1981 1982
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
1983 1984 1985 1986
{
	struct inode	*inode = dentry->d_inode;
	int		err;

1987
	if ((acc & NFSD_MAY_MASK) == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
1988 1989 1990 1991
		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
1992 1993 1994 1995 1996 1997 1998
		(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
1999 2000 2001
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2002
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2003
	dprintk("      owner %d/%d user %d/%d\n",
2004
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2005 2006 2007 2008 2009 2010
#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
2011 2012
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2013 2014
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2015
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2016
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2017 2018
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2019
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2020 2021
		return nfserr_perm;

M
Miklos Szeredi 已提交
2022
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2023 2024 2025 2026 2027 2028 2029
		/* 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
2030
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045
	}
	/*
	 * 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
2046
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2047
	    uid_eq(inode->i_uid, current_fsuid()))
L
Linus Torvalds 已提交
2048 2049
		return 0;

M
Miklos Szeredi 已提交
2050
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2051
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2052 2053 2054

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
2055 2056
	     (acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE) ||
	      acc == (NFSD_MAY_READ | NFSD_MAY_READ_IF_EXEC)))
2057
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
2058 2059 2060 2061 2062 2063 2064

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2065 2066 2067
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2068
	dprintk("nfsd: freeing readahead buffers.\n");
2069 2070 2071 2072 2073 2074 2075 2076 2077 2078

	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
2079 2080 2081 2082 2083 2084 2085 2086
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2087 2088
	int	j = 0;
	int	nperbucket;
2089
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2090

2091

2092
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2093
		return 0;
2094
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
2095
	nperbucket = max(2, nperbucket);
2096
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2097 2098

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2099 2100

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2101
		spin_lock_init(&raparm_hash[i].pb_lock);
2102 2103 2104 2105 2106 2107 2108 2109 2110

		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;
2111 2112
	}

L
Linus Torvalds 已提交
2113 2114
	nfsdstats.ra_size = cache_size;
	return 0;
2115 2116 2117 2118 2119

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2120
}