vfs.c 51 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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>
Anna Schumaker's avatar
Anna Schumaker committed
19
#include <linux/falloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22
#include <linux/fcntl.h>
#include <linux/namei.h>
#include <linux/delay.h>
Robert Love's avatar
Robert Love committed
23
#include <linux/fsnotify.h>
Linus Torvalds's avatar
Linus Torvalds committed
24 25
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
26 27
#include <linux/jhash.h>
#include <linux/ima.h>
28
#include <linux/slab.h>
29
#include <asm/uaccess.h>
30 31
#include <linux/exportfs.h>
#include <linux/writeback.h>
32
#include <linux/security.h>
33 34 35 36 37

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

38
#ifdef CONFIG_NFSD_V4
39 40
#include "acl.h"
#include "idmap.h"
Linus Torvalds's avatar
Linus Torvalds committed
41 42
#endif /* CONFIG_NFSD_V4 */

43 44
#include "nfsd.h"
#include "vfs.h"
Linus Torvalds's avatar
Linus Torvalds committed
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62

#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;
63
	unsigned int		p_hindex;
Linus Torvalds's avatar
Linus Torvalds committed
64 65
};

66 67 68 69 70 71 72 73 74
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];
Linus Torvalds's avatar
Linus Torvalds committed
75 76 77 78

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
79
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
Linus Torvalds's avatar
Linus Torvalds committed
80 81 82 83 84 85 86 87
 *  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;
Al Viro's avatar
Al Viro committed
88 89
	struct path path = {.mnt = mntget(exp->ex_path.mnt),
			    .dentry = dget(dentry)};
90
	int err = 0;
Linus Torvalds's avatar
Linus Torvalds committed
91

92
	err = follow_down(&path);
93 94
	if (err < 0)
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
95

Al Viro's avatar
Al Viro committed
96
	exp2 = rqst_exp_get_by_name(rqstp, &path);
Linus Torvalds's avatar
Linus Torvalds committed
97
	if (IS_ERR(exp2)) {
98 99 100 101 102 103 104 105 106 107
		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;
Al Viro's avatar
Al Viro committed
108
		path_put(&path);
Linus Torvalds's avatar
Linus Torvalds committed
109 110
		goto out;
	}
111 112
	if (nfsd_v4client(rqstp) ||
		(exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
Linus Torvalds's avatar
Linus Torvalds committed
113
		/* successfully crossed mount point */
Al Viro's avatar
Al Viro committed
114
		/*
Al Viro's avatar
Al Viro committed
115 116 117 118
		 * 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
Al Viro's avatar
Al Viro committed
119
		 */
Al Viro's avatar
Al Viro committed
120 121
		*dpp = path.dentry;
		path.dentry = dentry;
Al Viro's avatar
Al Viro committed
122
		*expp = exp2;
Al Viro's avatar
Al Viro committed
123
		exp2 = exp;
Linus Torvalds's avatar
Linus Torvalds committed
124
	}
Al Viro's avatar
Al Viro committed
125 126
	path_put(&path);
	exp_put(exp2);
Linus Torvalds's avatar
Linus Torvalds committed
127 128 129 130
out:
	return err;
}

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164
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;
}

165 166 167 168
/*
 * For nfsd purposes, we treat V4ROOT exports as though there was an
 * export at *every* directory.
 */
169
int nfsd_mountpoint(struct dentry *dentry, struct svc_export *exp)
170 171 172
{
	if (d_mountpoint(dentry))
		return 1;
173 174
	if (nfsd4_is_junction(dentry))
		return 1;
175 176
	if (!(exp->ex_flags & NFSEXP_V4ROOT))
		return 0;
177
	return d_inode(dentry) != NULL;
178 179
}

180
__be32
181
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
182
		   const char *name, unsigned int len,
183
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
Linus Torvalds's avatar
Linus Torvalds committed
184 185 186 187
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
188
	int			host_err;
Linus Torvalds's avatar
Linus Torvalds committed
189 190 191 192

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

	dparent = fhp->fh_dentry;
193
	exp = exp_get(fhp->fh_export);
Linus Torvalds's avatar
Linus Torvalds committed
194 195 196 197 198

	/* 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)
Linus Torvalds's avatar
Linus Torvalds committed
200
			dentry = dget_parent(dparent);
201
		else if (!EX_NOHIDE(exp) && !nfsd_v4client(rqstp))
Linus Torvalds's avatar
Linus Torvalds committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
216
		dentry = lookup_one_len(name, dparent, len);
217
		host_err = PTR_ERR(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
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))) {
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
268 269 270 271 272
	/*
	 * 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);
273
	if (!err && d_really_is_negative(dentry))
Linus Torvalds's avatar
Linus Torvalds committed
274
		err = nfserr_noent;
275
out:
Linus Torvalds's avatar
Linus Torvalds committed
276 277 278 279 280
	dput(dentry);
	exp_put(exp);
	return err;
}

281 282 283 284 285 286
/*
 * Commit metadata changes to stable storage.
 */
static int
commit_metadata(struct svc_fh *fhp)
{
287
	struct inode *inode = d_inode(fhp->fh_dentry);
288 289 290 291 292
	const struct export_operations *export_ops = inode->i_sb->s_export_op;

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

293 294 295
	if (export_ops->commit_metadata)
		return export_ops->commit_metadata(inode);
	return sync_inode_metadata(inode, 1);
296
}
297

Linus Torvalds's avatar
Linus Torvalds committed
298
/*
299 300
 * Go over the attributes and take care of the small differences between
 * NFS semantics and what Linux expects.
Linus Torvalds's avatar
Linus Torvalds committed
301
 */
302 303
static void
nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
Linus Torvalds's avatar
Linus Torvalds committed
304
{
305 306
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
Linus Torvalds's avatar
Linus Torvalds committed
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
314
	 * it is not an interface that we should be using.
Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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;
		}
Linus Torvalds's avatar
Linus Torvalds committed
339 340
	}

341
	/* sanitize the mode change */
Linus Torvalds's avatar
Linus Torvalds committed
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
		}
Linus Torvalds's avatar
Linus Torvalds committed
360
	}
361 362 363 364 365 366
}

static __be32
nfsd_get_write_access(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct iattr *iap)
{
367
	struct inode *inode = d_inode(fhp->fh_dentry);
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
	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;
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;

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);
	}
427 428

	dentry = fhp->fh_dentry;
429
	inode = d_inode(dentry);
430 431 432 433 434 435 436

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

	if (!iap->ia_valid)
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
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;
449 450 451 452 453 454 455 456 457 458

		/*
		 * RFC5661, Section 18.30.4:
		 *   Changing the size of a file with SETATTR indirectly
		 *   changes the time_modify and change attributes.
		 *
		 * (and similar for the older RFCs)
		 */
		if (iap->ia_size != i_size_read(inode))
			iap->ia_valid |= ATTR_MTIME;
459
	}
Linus Torvalds's avatar
Linus Torvalds committed
460 461 462

	iap->ia_valid |= ATTR_CTIME;

463 464 465
	if (check_guard && guardtime != inode->i_ctime.tv_sec) {
		err = nfserr_notsync;
		goto out_put_write_access;
Linus Torvalds's avatar
Linus Torvalds committed
466
	}
467 468 469 470

	fh_lock(fhp);
	host_err = notify_change(dentry, iap, NULL);
	fh_unlock(fhp);
471
	err = nfserrno(host_err);
472 473

out_put_write_access:
Linus Torvalds's avatar
Linus Torvalds committed
474 475 476
	if (size_change)
		put_write_access(inode);
	if (!err)
477
		err = nfserrno(commit_metadata(fhp));
Linus Torvalds's avatar
Linus Torvalds committed
478 479 480 481
out:
	return err;
}

482
#if defined(CONFIG_NFSD_V4)
483 484 485 486 487 488 489 490 491 492 493 494 495
/*
 * 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.
 */
496 497
int nfsd4_is_junction(struct dentry *dentry)
{
498
	struct inode *inode = d_inode(dentry);
499 500 501 502 503 504 505

	if (inode == NULL)
		return 0;
	if (inode->i_mode & S_IXUGO)
		return 0;
	if (!(inode->i_mode & S_ISVTX))
		return 0;
506
	if (vfs_getxattr(dentry, NFSD_JUNCTION_XATTR_NAME, NULL, 0) <= 0)
507 508 509
		return 0;
	return 1;
}
510 511 512 513 514 515 516 517 518 519 520 521 522 523
#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;

524
	mutex_lock(&d_inode(dentry)->i_mutex);
525
	host_error = security_inode_setsecctx(dentry, label->data, label->len);
526
	mutex_unlock(&d_inode(dentry)->i_mutex);
527 528 529 530 531 532 533 534 535 536
	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

Anna Schumaker's avatar
Anna Schumaker committed
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556
__be32 nfsd4_vfs_fallocate(struct svc_rqst *rqstp, struct svc_fh *fhp,
			   struct file *file, loff_t offset, loff_t len,
			   int flags)
{
	__be32 err;
	int error;

	if (!S_ISREG(file_inode(file)->i_mode))
		return nfserr_inval;

	err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry, NFSD_MAY_WRITE);
	if (err)
		return err;

	error = vfs_fallocate(file, flags, offset, len);
	if (!error)
		error = commit_metadata(fhp);

	return nfserrno(error);
}
J. Bruce Fields's avatar
J. Bruce Fields committed
557
#endif /* defined(CONFIG_NFSD_V4) */
Linus Torvalds's avatar
Linus Torvalds committed
558 559 560 561 562 563 564 565 566 567

#ifdef CONFIG_NFSD_V3
/*
 * Check server access rights to a file system object
 */
struct accessmap {
	u32		access;
	int		how;
};
static struct accessmap	nfs3_regaccess[] = {
Miklos Szeredi's avatar
Miklos Szeredi committed
568 569 570 571
    {	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			},
Linus Torvalds's avatar
Linus Torvalds committed
572 573 574 575 576

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
Miklos Szeredi's avatar
Miklos Szeredi committed
577 578 579 580 581
    {	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			},
Linus Torvalds's avatar
Linus Torvalds committed
582 583 584 585 586 587 588 589 590 591 592 593

    {	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
	 */
Miklos Szeredi's avatar
Miklos Szeredi committed
594 595 596 597
    {	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	},
Linus Torvalds's avatar
Linus Torvalds committed
598 599 600 601

    {	0,			0				}
};

602
__be32
Linus Torvalds's avatar
Linus Torvalds committed
603 604 605 606 607 608
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;
609
	__be32			error;
Linus Torvalds's avatar
Linus Torvalds committed
610

Miklos Szeredi's avatar
Miklos Szeredi committed
611
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
Linus Torvalds's avatar
Linus Torvalds committed
612 613 614 615 616 617
	if (error)
		goto out;

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

618
	if (d_is_reg(dentry))
Linus Torvalds's avatar
Linus Torvalds committed
619
		map = nfs3_regaccess;
620
	else if (d_is_dir(dentry))
Linus Torvalds's avatar
Linus Torvalds committed
621 622 623 624 625 626 627 628
		map = nfs3_diraccess;
	else
		map = nfs3_anyaccess;


	query = *access;
	for  (; map->access; map++) {
		if (map->access & query) {
629
			__be32 err2;
Linus Torvalds's avatar
Linus Torvalds committed
630 631 632

			sresult |= map->access;

633
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
Linus Torvalds's avatar
Linus Torvalds committed
634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
			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 */

661 662 663
static int nfsd_open_break_lease(struct inode *inode, int access)
{
	unsigned int mode;
Linus Torvalds's avatar
Linus Torvalds committed
664

665 666 667 668 669
	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);
}
Linus Torvalds's avatar
Linus Torvalds committed
670 671 672

/*
 * Open an existing file or directory.
673 674
 * The may_flags argument indicates the type of open (read/write/lock)
 * and additional flags.
Linus Torvalds's avatar
Linus Torvalds committed
675 676
 * N.B. After this call fhp needs an fh_put
 */
677
__be32
Al Viro's avatar
Al Viro committed
678
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
679
			int may_flags, struct file **filp)
Linus Torvalds's avatar
Linus Torvalds committed
680
{
681
	struct path	path;
Linus Torvalds's avatar
Linus Torvalds committed
682
	struct inode	*inode;
683
	struct file	*file;
684 685
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
686
	int		host_err = 0;
Linus Torvalds's avatar
Linus Torvalds committed
687

688 689
	validate_process_creds();

Linus Torvalds's avatar
Linus Torvalds committed
690 691 692 693
	/*
	 * If we get here, then the client has already done an "open",
	 * and (hopefully) checked permission - so allow OWNER_OVERRIDE
	 * in case a chmod has now revoked permission.
694 695 696 697 698 699
	 *
	 * 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.
Linus Torvalds's avatar
Linus Torvalds committed
700
	 */
701 702 703
	if (type == S_IFREG)
		may_flags |= NFSD_MAY_OWNER_OVERRIDE;
	err = fh_verify(rqstp, fhp, type, may_flags);
Linus Torvalds's avatar
Linus Torvalds committed
704 705 706
	if (err)
		goto out;

707 708
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
709
	inode = d_inode(path.dentry);
Linus Torvalds's avatar
Linus Torvalds committed
710 711 712 713 714

	/* Disallow write access to files with the append-only bit set
	 * or any access when mandatory locking enabled
	 */
	err = nfserr_perm;
715
	if (IS_APPEND(inode) && (may_flags & NFSD_MAY_WRITE))
Linus Torvalds's avatar
Linus Torvalds committed
716
		goto out;
717 718 719 720 721 722
	/*
	 * 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))
Linus Torvalds's avatar
Linus Torvalds committed
723 724 725 726 727
		goto out;

	if (!inode->i_fop)
		goto out;

728
	host_err = nfsd_open_break_lease(inode, may_flags);
729
	if (host_err) /* NOMEM or WOULDBLOCK */
Linus Torvalds's avatar
Linus Torvalds committed
730 731
		goto out_nfserr;

732 733
	if (may_flags & NFSD_MAY_WRITE) {
		if (may_flags & NFSD_MAY_READ)
734 735 736
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
Linus Torvalds's avatar
Linus Torvalds committed
737
	}
738

739 740 741 742 743 744
	file = dentry_open(&path, flags, current_cred());
	if (IS_ERR(file)) {
		host_err = PTR_ERR(file);
		goto out_nfserr;
	}

745
	host_err = ima_file_check(file, may_flags, 0);
746
	if (host_err) {
747
		fput(file);
748
		goto out_nfserr;
749 750
	}

751 752 753 754 755 756
	if (may_flags & NFSD_MAY_64BIT_COOKIE)
		file->f_mode |= FMODE_64BITHASH;
	else
		file->f_mode |= FMODE_32BITHASH;

	*filp = file;
Linus Torvalds's avatar
Linus Torvalds committed
757
out_nfserr:
758
	err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
759
out:
760
	validate_process_creds();
Linus Torvalds's avatar
Linus Torvalds committed
761 762 763 764 765 766 767 768 769 770 771 772 773
	return err;
}

/*
 * 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;
774 775 776 777 778
	unsigned int hash;
	struct raparm_hbucket *rab;

	hash = jhash_2words(dev, ino, 0xfeedbeef) & RAPARM_HASH_MASK;
	rab = &raparm_hash[hash];
Linus Torvalds's avatar
Linus Torvalds committed
779

780 781
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
Linus Torvalds's avatar
Linus Torvalds committed
782 783 784 785 786 787
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
788
	depth = nfsdstats.ra_size;
Linus Torvalds's avatar
Linus Torvalds committed
789
	if (!frap) {	
790
		spin_unlock(&rab->pb_lock);
Linus Torvalds's avatar
Linus Torvalds committed
791 792 793 794 795 796 797
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
798
	ra->p_hindex = hash;
Linus Torvalds's avatar
Linus Torvalds committed
799
found:
800
	if (rap != &rab->pb_head) {
Linus Torvalds's avatar
Linus Torvalds committed
801
		*rap = ra->p_next;
802 803
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
Linus Torvalds's avatar
Linus Torvalds committed
804 805 806
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
807
	spin_unlock(&rab->pb_lock);
Linus Torvalds's avatar
Linus Torvalds committed
808 809 810 811
	return ra;
}

/*
812 813 814
 * 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.
Linus Torvalds's avatar
Linus Torvalds committed
815 816
 */
static int
817 818
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
Linus Torvalds's avatar
Linus Torvalds committed
819
{
820
	struct svc_rqst *rqstp = sd->u.data;
821
	struct page **pp = rqstp->rq_next_page;
822 823 824 825
	struct page *page = buf->page;
	size_t size;

	size = sd->len;
Linus Torvalds's avatar
Linus Torvalds committed
826 827 828

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
829 830
		put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
831
		rqstp->rq_res.page_base = buf->offset;
Linus Torvalds's avatar
Linus Torvalds committed
832
		rqstp->rq_res.page_len = size;
833
	} else if (page != pp[-1]) {
Linus Torvalds's avatar
Linus Torvalds committed
834
		get_page(page);
835 836 837
		if (*rqstp->rq_next_page)
			put_page(*rqstp->rq_next_page);
		*(rqstp->rq_next_page++) = page;
Linus Torvalds's avatar
Linus Torvalds committed
838
		rqstp->rq_res.page_len += size;
839
	} else
Linus Torvalds's avatar
Linus Torvalds committed
840 841 842 843 844
		rqstp->rq_res.page_len += size;

	return size;
}

845 846 847 848 849 850
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

851 852
static __be32
nfsd_finish_read(struct file *file, unsigned long *count, int host_err)
Linus Torvalds's avatar
Linus Torvalds committed
853
{
854 855 856
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
857
		fsnotify_access(file);
858
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
859
	} else 
860 861 862
		return nfserrno(host_err);
}

863
__be32 nfsd_splice_read(struct svc_rqst *rqstp,
864 865 866 867 868 869 870 871 872 873 874 875 876 877 878
		     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);
}

879
__be32 nfsd_readv(struct file *file, loff_t offset, struct kvec *vec, int vlen,
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895
		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)
{
896
	if (file->f_op->splice_read && test_bit(RQ_SPLICE_OK, &rqstp->rq_flags))
897 898 899
		return nfsd_splice_read(rqstp, file, offset, count);
	else
		return nfsd_readv(file, offset, vec, vlen, count);
Linus Torvalds's avatar
Linus Torvalds committed
900 901
}

902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917
/*
 * 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)
{
Al Viro's avatar
Al Viro committed
918
	struct inode *inode = file_inode(file);
919 920 921 922 923 924 925 926 927 928 929 930 931
	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));
932
		err = vfs_fsync(file, 0);
933 934 935 936 937 938
	}
	last_ino = inode->i_ino;
	last_dev = inode->i_sb->s_dev;
	return err;
}

939
static __be32
Linus Torvalds's avatar
Linus Torvalds committed
940 941
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
942
				unsigned long *cnt, int *stablep)
Linus Torvalds's avatar
Linus Torvalds committed
943 944 945 946
{
	struct svc_export	*exp;
	struct inode		*inode;
	mm_segment_t		oldfs;
947 948
	__be32			err = 0;
	int			host_err;
Linus Torvalds's avatar
Linus Torvalds committed
949
	int			stable = *stablep;
950
	int			use_wgather;
Kent Overstreet's avatar
Kent Overstreet committed
951
	loff_t			pos = offset;
952
	loff_t			end = LLONG_MAX;
953 954
	unsigned int		pflags = current->flags;

955
	if (test_bit(RQ_LOCAL, &rqstp->rq_flags))
956 957 958 959 960 961 962
		/*
		 * 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;
Linus Torvalds's avatar
Linus Torvalds committed
963

Al Viro's avatar
Al Viro committed
964
	inode = file_inode(file);
Linus Torvalds's avatar
Linus Torvalds committed
965 966
	exp   = fhp->fh_export;

967
	use_wgather = (rqstp->rq_vers == 2) && EX_WGATHER(exp);
Linus Torvalds's avatar
Linus Torvalds committed
968 969 970 971 972 973

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

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
Kent Overstreet's avatar
Kent Overstreet committed
974
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &pos);
Linus Torvalds's avatar
Linus Torvalds committed
975
	set_fs(oldfs);
976 977 978 979
	if (host_err < 0)
		goto out_nfserr;
	*cnt = host_err;
	nfsdstats.io_write += host_err;
980
	fsnotify_modify(file);
Linus Torvalds's avatar
Linus Torvalds committed
981

982
	if (stable) {
983
		if (use_wgather) {
984
			host_err = wait_for_concurrent_writes(file);
985 986 987 988 989
		} else {
			if (*cnt)
				end = offset + *cnt - 1;
			host_err = vfs_fsync_range(file, offset, end, 0);
		}
990
	}
Linus Torvalds's avatar
Linus Torvalds committed
991

992
out_nfserr:
993
	dprintk("nfsd: write complete host_err=%d\n", host_err);
994
	if (host_err >= 0)
Linus Torvalds's avatar
Linus Torvalds committed
995
		err = 0;
996
	else
997
		err = nfserrno(host_err);
998
	if (test_bit(RQ_LOCAL, &rqstp->rq_flags))
999
		tsk_restore_flags(current, pflags, PF_LESS_THROTTLE);
Linus Torvalds's avatar
Linus Torvalds committed
1000 1001 1002
	return err;
}

1003 1004
__be32 nfsd_get_tmp_read_open(struct svc_rqst *rqstp, struct svc_fh *fhp,
		struct file **file, struct raparms **ra)
1005 1006 1007 1008
{
	struct inode *inode;
	__be32 err;

1009
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, file);
1010 1011 1012
	if (err)
		return err;

1013
	inode = file_inode(*file);
1014 1015

	/* Get readahead parameters */
1016
	*ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);
1017

1018 1019 1020 1021
	if (*ra && (*ra)->p_set)
		(*file)->f_ra = (*ra)->p_ra;
	return nfs_ok;
}
1022

1023 1024
void nfsd_put_tmp_read_open(struct file *file, struct raparms *ra)
{
1025 1026 1027 1028 1029 1030 1031 1032 1033
	/* 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);
	}
1034
	fput(file);
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
}

/*
 * 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);

1057 1058 1059
	return err;
}

Linus Torvalds's avatar
Linus Torvalds committed
1060 1061 1062 1063 1064
/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1065
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1066
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1067
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
Linus Torvalds's avatar
Linus Torvalds committed
1068 1069
		int *stablep)
{
1070
	__be32			err = 0;
Linus Torvalds's avatar
Linus Torvalds committed
1071 1072

	if (file) {
1073
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
Miklos Szeredi's avatar
Miklos Szeredi committed
1074
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
Linus Torvalds's avatar
Linus Torvalds committed
1075 1076 1077 1078 1079
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
Miklos Szeredi's avatar
Miklos Szeredi committed
1080
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
Linus Torvalds's avatar
Linus Torvalds committed
1081 1082 1083 1084 1085 1086
		if (err)
			goto out;

		if (cnt)
			err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen,
					     cnt, stablep);
1087
		fput(file);
Linus Torvalds's avatar
Linus Torvalds committed
1088 1089 1090 1091 1092 1093 1094 1095
	}
out:
	return err;
}

#ifdef CONFIG_NFSD_V3
/*
 * Commit all pending writes to stable storage.
1096 1097 1098
 *
 * Note: we only guarantee that data that lies within the range specified
 * by the 'offset' and 'count' parameters will be synced.
Linus Torvalds's avatar
Linus Torvalds committed
1099 1100 1101 1102
 *
 * 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.
 */
1103
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1104 1105 1106 1107
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1108 1109
	loff_t		end = LLONG_MAX;
	__be32		err = nfserr_inval;
Linus Torvalds's avatar
Linus Torvalds committed
1110

1111 1112 1113 1114 1115 1116 1117
	if (offset < 0)
		goto out;
	if (count != 0) {
		end = offset + (loff_t)count - 1;
		if (end < offset)
			goto out;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1118

1119 1120
	err = nfsd_open(rqstp, fhp, S_IFREG,
			NFSD_MAY_WRITE|NFSD_MAY_NOT_BREAK_LEASE, &file);
Miklos Szeredi's avatar
Miklos Szeredi committed
1121
	if (err)
1122
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
1123
	if (EX_ISSYNC(fhp->fh_export)) {
1124
		int err2 = vfs_fsync_range(file, offset, end, 0);
1125 1126 1127 1128

		if (err2 != -EINVAL)
			err = nfserrno(err2);
		else
Linus Torvalds's avatar
Linus Torvalds committed
1129 1130 1131
			err = nfserr_notsupp;
	}

1132
	fput(file);
1133
out:
Linus Torvalds's avatar
Linus Torvalds committed
1134 1135 1136 1137
	return err;
}
#endif /* CONFIG_NFSD_V3 */

1138
static __be32
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
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:
	 */
1151
	if (!uid_eq(current_fsuid(), GLOBAL_ROOT_UID))
1152 1153 1154
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
1155
	/* Callers expect file metadata to be committed here */
1156
	return nfserrno(commit_metadata(resfhp));
1157 1158
}

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
/* 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;
}

Linus Torvalds's avatar
Linus Torvalds committed
1174 1175 1176 1177 1178 1179 1180 1181
/*
 * 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
 */
1182
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1183 1184 1185 1186 1187 1188
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;
1189
	__be32		err;
1190
	__be32		err2;
1191
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1192 1193 1194 1195 1196 1197 1198 1199

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

Miklos Szeredi's avatar
Miklos Szeredi committed
1200
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
Linus Torvalds's avatar
Linus Torvalds committed
1201 1202 1203 1204
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
1205
	dirp = d_inode(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
1206 1207

	err = nfserr_notdir;
Al Viro's avatar
Al Viro committed
1208
	if (!dirp->i_op->lookup)
Linus Torvalds's avatar
Linus Torvalds committed
1209 1210 1211 1212 1213 1214
		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) {
1215 1216 1217 1218
		host_err = fh_want_write(fhp);
		if (host_err)
			goto out_nfserr;

Linus Torvalds's avatar
Linus Torvalds committed
1219
		/* called from nfsd_proc_mkdir, or possibly nfsd3_proc_create */
1220
		fh_lock_nested(fhp, I_MUTEX_PARENT);
Linus Torvalds's avatar
Linus Torvalds committed
1221
		dchild = lookup_one_len(fname, dentry, flen);
1222
		host_err = PTR_ERR(dchild);
Linus Torvalds's avatar
Linus Torvalds committed
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
		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
Al Viro's avatar
Al Viro committed
1234 1235
				"nfsd_create: parent %pd2 not locked!\n",
				dentry);
1236
			err = nfserr_io;
Linus Torvalds's avatar
Linus Torvalds committed
1237 1238 1239 1240 1241 1242 1243
			goto out;
		}
	}
	/*
	 * Make sure the child dentry is still negative ...
	 */
	err = nfserr_exist;
1244
	if (d_really_is_positive(dchild)) {
Al Viro's avatar
Al Viro committed
1245 1246
		dprintk("nfsd_create: dentry %pd/%pd not negative!\n",
			dentry, dchild);
Linus Torvalds's avatar
Linus Torvalds committed
1247 1248 1249 1250 1251 1252 1253
		goto out; 
	}

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

1254 1255 1256 1257 1258 1259 1260
	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;
	}

Linus Torvalds's avatar
Linus Torvalds committed
1261 1262 1263
	/*
	 * Get the dir op function pointer.
	 */
1264
	err = 0;
1265
	host_err = 0;
Linus Torvalds's avatar
Linus Torvalds committed
1266 1267
	switch (type) {
	case S_IFREG:
1268
		host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1269 1270
		if (!host_err)
			nfsd_check_ignore_resizing(iap);
Linus Torvalds's avatar
Linus Torvalds committed
1271 1272
		break;
	case S_IFDIR:
1273
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
Linus Torvalds's avatar
Linus Torvalds committed
1274 1275 1276 1277 1278
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1279
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
Linus Torvalds's avatar
Linus Torvalds committed
1280 1281
		break;
	}
1282
	if (host_err < 0)
Linus Torvalds's avatar
Linus Torvalds committed
1283 1284
		goto out_nfserr;

1285
	err = nfsd_create_setattr(rqstp, resfhp, iap);
Linus Torvalds's avatar
Linus Torvalds committed
1286

1287
	/*
1288 1289 1290 1291
	 * nfsd_create_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.
1292 1293
	 */
	err2 = nfserrno(commit_metadata(fhp));
1294 1295
	if (err2)
		err = err2;
Linus Torvalds's avatar
Linus Torvalds committed
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
	/*
	 * 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:
1307
	err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
1308 1309 1310 1311
	goto out;
}

#ifdef CONFIG_NFSD_V3
1312 1313 1314 1315 1316 1317 1318

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

Linus Torvalds's avatar
Linus Torvalds committed
1319
/*
1320
 * NFSv3 and NFSv4 version of nfsd_create
Linus Torvalds's avatar
Linus Torvalds committed
1321
 */
1322
__be32
1323
do_nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
Linus Torvalds's avatar
Linus Torvalds committed
1324 1325
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1326
	        bool *truncp, bool *created)
Linus Torvalds's avatar
Linus Torvalds committed
1327 1328 1329
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1330 1331
	__be32		err;
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
	__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;
1342
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
1343 1344 1345 1346
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
1347
	dirp = d_inode(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
1348 1349 1350 1351

	/* Get all the sanity checks out of the way before
	 * we lock the parent. */
	err = nfserr_notdir;
Al Viro's avatar
Al Viro committed
1352
	if (!dirp->i_op->lookup)
Linus Torvalds's avatar
Linus Torvalds committed
1353
		goto out;
1354 1355 1356 1357 1358

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

1359
	fh_lock_nested(fhp, I_MUTEX_PARENT);
Linus Torvalds's avatar
Linus Torvalds committed
1360 1361 1362 1363 1364

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1365
	host_err = PTR_ERR(dchild);
Linus Torvalds's avatar
Linus Torvalds committed
1366 1367 1368
	if (IS_ERR(dchild))
		goto out_nfserr;

1369
	/* If file doesn't exist, check for permissions to create one */
1370
	if (d_really_is_negative(dchild)) {
1371 1372 1373 1374 1375
		err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
		if (err)
			goto out;
	}

Linus Torvalds's avatar
Linus Torvalds committed
1376 1377 1378 1379
	err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
	if (err)
		goto out;

1380
	if (nfsd_create_is_exclusive(createmode)) {
1381
		/* solaris7 gets confused (bugid 4218508) if these have
1382 1383 1384 1385
		 * 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.
Linus Torvalds's avatar
Linus Torvalds committed
1386 1387 1388 1389 1390
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1391
	if (d_really_is_positive(dchild)) {
Linus Torvalds's avatar
Linus Torvalds committed
1392 1393 1394 1395
		err = 0;

		switch (createmode) {
		case NFS3_CREATE_UNCHECKED:
1396
			if (! d_is_reg(dchild))
1397
				goto out;
Linus Torvalds's avatar
Linus Torvalds committed
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
			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:
1414 1415 1416
			if (   d_inode(dchild)->i_mtime.tv_sec == v_mtime
			    && d_inode(dchild)->i_atime.tv_sec == v_atime
			    && d_inode(dchild)->i_size  == 0 ) {
1417 1418
				if (created)
					*created = 1;
Linus Torvalds's avatar
Linus Torvalds committed
1419
				break;
1420
			}
1421
		case NFS4_CREATE_EXCLUSIVE4_1:
1422 1423 1424
			if (   d_inode(dchild)->i_mtime.tv_sec == v_mtime
			    && d_inode(dchild)->i_atime.tv_sec == v_atime
			    && d_inode(dchild)->i_size  == 0 ) {
1425 1426
				if (created)
					*created = 1;
1427
				goto set_attr;
1428
			}
Linus Torvalds's avatar
Linus Torvalds committed
1429 1430 1431 1432
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
1433
		fh_drop_write(fhp);
Linus Torvalds's avatar
Linus Torvalds committed
1434 1435 1436
		goto out;
	}

1437
	host_err = vfs_create(dirp, dchild, iap->ia_mode, true);
1438
	if (host_err < 0) {
1439
		fh_drop_write(fhp);
Linus Torvalds's avatar
Linus Torvalds committed
1440
		goto out_nfserr;
1441
	}
1442 1443
	if (created)
		*created = 1;
Linus Torvalds's avatar
Linus Torvalds committed
1444

1445 1446
	nfsd_check_ignore_resizing(iap);

1447
	if (nfsd_create_is_exclusive(createmode)) {
1448
		/* Cram the verifier into atime/mtime */
Linus Torvalds's avatar
Linus Torvalds committed
1449
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1450
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
Linus Torvalds's avatar
Linus Torvalds committed
1451 1452 1453 1454 1455 1456 1457 1458
		/* 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:
1459 1460 1461
	err = nfsd_create_setattr(rqstp, resfhp, iap);

	/*
1462 1463
	 * nfsd_create_setattr already committed the child
	 * (and possibly also the parent).
1464 1465 1466
	 */
	if (!err)
		err = nfserrno(commit_metadata(fhp));
1467 1468 1469 1470 1471 1472

	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
Linus Torvalds's avatar
Linus Torvalds committed
1473 1474 1475 1476 1477

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
1478
	fh_drop_write(fhp);
Linus Torvalds's avatar
Linus Torvalds committed
1479 1480 1481
 	return err;
 
 out_nfserr:
1482
	err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
1483 1484 1485 1486 1487 1488 1489 1490 1491
	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
 */
1492
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1493 1494 1495 1496
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct inode	*inode;
	mm_segment_t	oldfs;
1497 1498
	__be32		err;
	int		host_err;
Al Viro's avatar
Al Viro committed
1499
	struct path path;
Linus Torvalds's avatar
Linus Torvalds committed
1500

Miklos Szeredi's avatar
Miklos Szeredi committed
1501
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
Linus Torvalds's avatar
Linus Torvalds committed
1502 1503 1504
	if (err)
		goto out;

Al Viro's avatar
Al Viro committed
1505 1506
	path.mnt = fhp->fh_export->ex_path.mnt;
	path.dentry = fhp->fh_dentry;
1507
	inode = d_inode(path.dentry);
Linus Torvalds's avatar
Linus Torvalds committed
1508 1509

	err = nfserr_inval;
Al Viro's avatar
Al Viro committed
1510
	if (!inode->i_op->readlink)
Linus Torvalds's avatar
Linus Torvalds committed
1511 1512
		goto out;

Al Viro's avatar
Al Viro committed
1513
	touch_atime(&path);
Linus Torvalds's avatar
Linus Torvalds committed
1514 1515 1516 1517 1518
	/* 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);
1519
	host_err = inode->i_op->readlink(path.dentry, (char __user *)buf, *lenp);
Linus Torvalds's avatar
Linus Torvalds committed
1520 1521
	set_fs(oldfs);

1522
	if (host_err < 0)
Linus Torvalds's avatar
Linus Torvalds committed
1523
		goto out_nfserr;
1524
	*lenp = host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1525 1526 1527 1528 1529
	err = 0;
out:
	return err;

out_nfserr:
1530
	err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
1531 1532 1533 1534 1535 1536 1537
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1538
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1539 1540
nfsd_symlink(struct svc_rqst *rqstp, struct svc_fh *fhp,
				char *fname, int flen,
1541
				char *path,
1542
				struct svc_fh *resfhp)
Linus Torvalds's avatar
Linus Torvalds committed
1543 1544
{
	struct dentry	*dentry, *dnew;
1545 1546
	__be32		err, cerr;
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1547 1548

	err = nfserr_noent;
1549
	if (!flen || path[0] == '\0')
Linus Torvalds's avatar
Linus Torvalds committed
1550 1551 1552 1553 1554
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;

Miklos Szeredi's avatar
Miklos Szeredi committed
1555
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
Linus Torvalds's avatar
Linus Torvalds committed
1556 1557
	if (err)
		goto out;
1558 1559 1560 1561 1562

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

Linus Torvalds's avatar
Linus Torvalds committed
1563 1564 1565
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1566
	host_err = PTR_ERR(dnew);
Linus Torvalds's avatar
Linus Torvalds committed
1567 1568 1569
	if (IS_ERR(dnew))
		goto out_nfserr;

1570
	host_err = vfs_symlink(d_inode(dentry), dnew, path);
1571
	err = nfserrno(host_err);
1572 1573
	if (!err)
		err = nfserrno(commit_metadata(fhp));
Linus Torvalds's avatar
Linus Torvalds committed
1574 1575
	fh_unlock(fhp);

1576
	fh_drop_write(fhp);
1577

Linus Torvalds's avatar
Linus Torvalds committed
1578 1579 1580 1581 1582 1583 1584
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1585
	err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
1586 1587 1588 1589 1590 1591 1592
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1593
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1594 1595 1596 1597
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
1598
	struct inode	*dirp;
1599 1600
	__be32		err;
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1601

Miklos Szeredi's avatar
Miklos Szeredi committed
1602
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
Linus Torvalds's avatar
Linus Torvalds committed
1603 1604
	if (err)
		goto out;
1605
	err = fh_verify(rqstp, tfhp, 0, NFSD_MAY_NOP);
Linus Torvalds's avatar
Linus Torvalds committed
1606 1607
	if (err)
		goto out;
1608
	err = nfserr_isdir;
1609
	if (d_is_dir(tfhp->fh_dentry))
1610
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
1611 1612 1613 1614 1615 1616 1617
	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

1618 1619 1620 1621 1622 1623
	host_err = fh_want_write(tfhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

1624
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
Linus Torvalds's avatar
Linus Torvalds committed
1625
	ddir = ffhp->fh_dentry;
1626
	dirp = d_inode(ddir);
Linus Torvalds's avatar
Linus Torvalds committed
1627 1628

	dnew = lookup_one_len(name, ddir, len);
1629
	host_err = PTR_ERR(dnew);
Linus Torvalds's avatar
Linus Torvalds committed
1630 1631 1632 1633 1634
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;

1635
	err = nfserr_noent;
1636
	if (d_really_is_negative(dold))
1637
		goto out_dput;
1638
	host_err = vfs_link(dold, dirp, dnew, NULL);
1639
	if (!host_err) {
1640 1641 1642
		err = nfserrno(commit_metadata(ffhp));
		if (!err)
			err = nfserrno(commit_metadata(tfhp));
Linus Torvalds's avatar
Linus Torvalds committed
1643
	} else {
1644
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
Linus Torvalds's avatar
Linus Torvalds committed
1645 1646
			err = nfserr_acces;
		else
1647
			err = nfserrno(host_err);
Linus Torvalds's avatar
Linus Torvalds committed
1648
	}
1649
out_dput:
Linus Torvalds's avatar
Linus Torvalds committed
1650
	dput(dnew);
1651 1652
out_unlock:
	fh_unlock(ffhp);
1653
	fh_drop_write(tfhp);
Linus Torvalds's avatar
Linus Torvalds committed
1654 1655 1656 1657
out:
	return err;

out_nfserr:
1658
	err = nfserrno(host_err);
1659
	goto out_unlock;
Linus Torvalds's avatar
Linus Torvalds committed
1660 1661 1662 1663 1664 1665
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1666
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1667 1668 1669 1670 1671
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;
1672 1673
	__be32		err;
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1674

Miklos Szeredi's avatar
Miklos Szeredi committed
1675
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
Linus Torvalds's avatar
Linus Torvalds committed
1676 1677
	if (err)
		goto out;
Miklos Szeredi's avatar
Miklos Szeredi committed
1678
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
Linus Torvalds's avatar
Linus Torvalds committed
1679 1680 1681 1682
	if (err)
		goto out;

	fdentry = ffhp->fh_dentry;
1683
	fdir = d_inode(fdentry);
Linus Torvalds's avatar
Linus Torvalds committed
1684 1685

	tdentry = tfhp->fh_dentry;
1686
	tdir = d_inode(tdentry);
Linus Torvalds's avatar
Linus Torvalds committed
1687 1688 1689 1690 1691

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

1692 1693 1694 1695 1696 1697
	host_err = fh_want_write(ffhp);
	if (host_err) {
		err = nfserrno(host_err);
		goto out;
	}

Linus Torvalds's avatar
Linus Torvalds committed
1698 1699 1700 1701 1702 1703 1704 1705
	/* 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);
1706
	host_err = PTR_ERR(odentry);
Linus Torvalds's avatar
Linus Torvalds committed
1707 1708 1709
	if (IS_ERR(odentry))
		goto out_nfserr;

1710
	host_err = -ENOENT;
1711
	if (d_really_is_negative(odentry))
Linus Torvalds's avatar
Linus Torvalds committed
1712
		goto out_dput_old;
1713
	host_err = -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
1714 1715 1716 1717
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1718
	host_err = PTR_ERR(ndentry);
Linus Torvalds's avatar
Linus Torvalds committed
1719 1720
	if (IS_ERR(ndentry))
		goto out_dput_old;
1721
	host_err = -ENOTEMPTY;
Linus Torvalds's avatar
Linus Torvalds committed
1722 1723 1724
	if (ndentry == trap)
		goto out_dput_new;

1725 1726 1727
	host_err = -EXDEV;
	if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
		goto out_dput_new;
1728 1729
	if (ffhp->fh_export->ex_path.dentry != tfhp->fh_export->ex_path.dentry)
		goto out_dput_new;
1730

Miklos Szeredi's avatar
Miklos Szeredi committed
1731
	host_err = vfs_rename(fdir, odentry, tdir, ndentry, NULL, 0);
1732 1733
	if (!host_err) {
		host_err = commit_metadata(tfhp);
1734
		if (!host_err)
1735
			host_err = commit_metadata(ffhp);
Linus Torvalds's avatar
Linus Torvalds committed
1736 1737 1738 1739 1740 1741
	}
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1742
	err = nfserrno(host_err);
1743 1744
	/*
	 * We cannot rely on fh_unlock on the two filehandles,
Linus Torvalds's avatar
Linus Torvalds committed
1745
	 * as that would do the wrong thing if the two directories
1746
	 * were the same, so again we do it by hand.
Linus Torvalds's avatar
Linus Torvalds committed
1747 1748 1749 1750 1751
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;
1752
	fh_drop_write(ffhp);
Linus Torvalds's avatar
Linus Torvalds committed
1753 1754 1755 1756 1757 1758 1759 1760 1761

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
1762
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1763 1764 1765 1766 1767
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1768 1769
	__be32		err;
	int		host_err;
Linus Torvalds's avatar
Linus Torvalds committed
1770 1771 1772 1773

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
Miklos Szeredi's avatar
Miklos Szeredi committed
1774
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
Linus Torvalds's avatar
Linus Torvalds committed
1775 1776 1777
	if (err)
		goto out;

1778 1779 1780 1781
	host_err = fh_want_write(fhp);
	if (host_err)
		goto out_nfserr;

1782
	fh_lock_nested(fhp, I_MUTEX_PARENT);
Linus Torvalds's avatar
Linus Torvalds committed
1783
	dentry = fhp->fh_dentry;
1784
	dirp = d_inode(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
1785 1786

	rdentry = lookup_one_len(fname, dentry, flen);
1787
	host_err = PTR_ERR(rdentry);
Linus Torvalds's avatar
Linus Torvalds committed
1788 1789 1790
	if (IS_ERR(rdentry))
		goto out_nfserr;

1791
	if (d_really_is_negative(rdentry)) {
Linus Torvalds's avatar
Linus Torvalds committed
1792 1793 1794 1795 1796 1797
		dput(rdentry);
		err = nfserr_noent;
		goto out;
	}

	if (!type)
1798
		type = d_inode(rdentry)->i_mode & S_IFMT;
Linus Torvalds's avatar
Linus Torvalds committed
1799

1800
	if (type != S_IFDIR)
1801
		host_err = vfs_unlink(dirp, rdentry, NULL);
1802
	else
1803
		host_err = vfs_rmdir(dirp, rdentry);
1804 1805
	if (!host_err)
		host_err = commit_metadata(fhp);
1806 1807
	dput(rdentry);

Linus Torvalds's avatar
Linus Torvalds committed
1808
out_nfserr:
1809
	err = nfserrno(host_err);
1810 1811
out:
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
1812 1813
}

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

1836 1837 1838
static int nfsd_buffered_filldir(struct dir_context *ctx, const char *name,
				 int namlen, loff_t offset, u64 ino,
				 unsigned int d_type)
1839
{
1840 1841
	struct readdir_data *buf =
		container_of(ctx, struct readdir_data, ctx);
1842 1843 1844 1845
	struct buffered_dirent *de = (void *)(buf->dirent + buf->used);
	unsigned int reclen;

	reclen = ALIGN(sizeof(struct buffered_dirent) + namlen, sizeof(u64));
1846 1847
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1848
		return -EINVAL;
1849
	}
1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860

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

	return 0;
}

1861
static __be32 nfsd_buffered_readdir(struct file *file, nfsd_filldir_t func,
1862
				    struct readdir_cd *cdp, loff_t *offsetp)
1863
{
1864
	struct buffered_dirent *de;
1865
	int host_err;
1866 1867
	int size;
	loff_t offset;
Al Viro's avatar
Al Viro committed
1868 1869 1870 1871
	struct readdir_data buf = {
		.ctx.actor = nfsd_buffered_filldir,
		.dirent = (void *)__get_free_page(GFP_KERNEL)
	};
1872

1873
	if (!buf.dirent)
1874
		return nfserrno(-ENOMEM);
1875 1876

	offset = *offsetp;
1877

1878
	while (1) {
Al Viro's avatar
Al Viro committed
1879
		struct inode *dir_inode = file_inode(file);
1880 1881
		unsigned int reclen;

1882
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1883
		buf.used = 0;
1884
		buf.full = 0;
1885

1886
		host_err = iterate_dir(file, &buf.ctx);
1887 1888 1889 1890
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1891 1892 1893 1894 1895 1896 1897
			break;

		size = buf.used;

		if (!size)
			break;

1898 1899 1900 1901 1902 1903 1904 1905 1906
		/*
		 * 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;

1907 1908 1909 1910 1911 1912
		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))
1913
				break;
1914 1915

			if (cdp->err != nfs_ok)
1916
				break;
1917 1918 1919 1920 1921 1922

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1923 1924 1925 1926
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1927
		offset = vfs_llseek(file, 0, SEEK_CUR);
1928 1929 1930
	}

	free_page((unsigned long)(buf.dirent));
1931 1932 1933

	if (host_err)
		return nfserrno(host_err);
1934 1935 1936

	*offsetp = offset;
	return cdp->err;
1937 1938
}

Linus Torvalds's avatar
Linus Torvalds committed
1939 1940 1941 1942
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1943
__be32
Linus Torvalds's avatar
Linus Torvalds committed
1944
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1945
	     struct readdir_cd *cdp, nfsd_filldir_t func)
Linus Torvalds's avatar
Linus Torvalds committed
1946
{
1947
	__be32		err;
Linus Torvalds's avatar
Linus Torvalds committed
1948 1949
	struct file	*file;
	loff_t		offset = *offsetp;
1950 1951 1952 1953 1954
	int             may_flags = NFSD_MAY_READ;

	/* NFSv2 only supports 32 bit cookies */
	if (rqstp->rq_vers > 2)
		may_flags |= NFSD_MAY_64BIT_COOKIE;
Linus Torvalds's avatar
Linus Torvalds committed
1955

1956
	err = nfsd_open(rqstp, fhp, S_IFDIR, may_flags, &file);
Linus Torvalds's avatar
Linus Torvalds committed
1957 1958 1959
	if (err)
		goto out;

1960
	offset = vfs_llseek(file, offset, SEEK_SET);
Linus Torvalds's avatar
Linus Torvalds committed
1961 1962 1963 1964 1965
	if (offset < 0) {
		err = nfserrno((int)offset);
		goto out_close;
	}

1966
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
Linus Torvalds's avatar
Linus Torvalds committed
1967 1968 1969 1970

	if (err == nfserr_eof || err == nfserr_toosmall)
		err = nfs_ok; /* can still be found in ->err */
out_close:
1971
	fput(file);
Linus Torvalds's avatar
Linus Torvalds committed
1972 1973 1974 1975 1976 1977 1978 1979
out:
	return err;
}

/*
 * Get file system stats
 * N.B. After this call fhp needs an fh_put
 */
1980
__be32
1981
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
Linus Torvalds's avatar
Linus Torvalds committed
1982
{
1983 1984 1985
	__be32 err;

	err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
1986 1987 1988 1989 1990 1991 1992 1993
	if (!err) {
		struct path path = {
			.mnt	= fhp->fh_export->ex_path.mnt,
			.dentry	= fhp->fh_dentry,
		};
		if (vfs_statfs(&path, stat))
			err = nfserr_io;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1994 1995 1996
	return err;
}

J. Bruce Fields's avatar
J. Bruce Fields committed
1997
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
1998
{
J. Bruce Fields's avatar
J. Bruce Fields committed
1999
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
2000 2001
}

Linus Torvalds's avatar
Linus Torvalds committed
2002 2003 2004
/*
 * Check for a user's access permissions to this inode.
 */
2005
__be32
2006 2007
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
Linus Torvalds's avatar
Linus Torvalds committed
2008
{
2009
	struct inode	*inode = d_inode(dentry);
Linus Torvalds's avatar
Linus Torvalds committed
2010 2011
	int		err;

2012
	if ((acc & NFSD_MAY_MASK) == NFSD_MAY_NOP)
Linus Torvalds's avatar
Linus Torvalds committed
2013 2014 2015 2016
		return 0;
#if 0
	dprintk("nfsd: permission 0x%x%s%s%s%s%s%s%s mode 0%o%s%s%s\n",
		acc,
Miklos Szeredi's avatar
Miklos Szeredi committed
2017 2018 2019 2020 2021 2022 2023
		(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" : "",
Linus Torvalds's avatar
Linus Torvalds committed
2024 2025 2026
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2027
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
Linus Torvalds's avatar
Linus Torvalds committed
2028
	dprintk("      owner %d/%d user %d/%d\n",
2029
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
Linus Torvalds's avatar
Linus Torvalds committed
2030 2031 2032 2033 2034 2035
#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.
	 */
Miklos Szeredi's avatar
Miklos Szeredi committed
2036 2037
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2038 2039
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
Linus Torvalds's avatar
Linus Torvalds committed
2040
				return nfserr_rofs;
Miklos Szeredi's avatar
Miklos Szeredi committed
2041
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
Linus Torvalds's avatar
Linus Torvalds committed
2042 2043
				return nfserr_perm;
		}
Miklos Szeredi's avatar
Miklos Szeredi committed
2044
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
Linus Torvalds's avatar
Linus Torvalds committed
2045 2046
		return nfserr_perm;

Miklos Szeredi's avatar
Miklos Szeredi committed
2047
	if (acc & NFSD_MAY_LOCK) {
Linus Torvalds's avatar
Linus Torvalds committed
2048 2049 2050 2051 2052 2053 2054
		/* 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
Miklos Szeredi's avatar
Miklos Szeredi committed
2055
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
Linus Torvalds's avatar
Linus Torvalds committed
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070
	}
	/*
	 * 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.
	 */
Miklos Szeredi's avatar
Miklos Szeredi committed
2071
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2072
	    uid_eq(inode->i_uid, current_fsuid()))
Linus Torvalds's avatar
Linus Torvalds committed
2073 2074
		return 0;

Miklos Szeredi's avatar
Miklos Szeredi committed
2075
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2076
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
Linus Torvalds's avatar
Linus Torvalds committed
2077 2078 2079

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
2080 2081
	     (acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE) ||
	      acc == (NFSD_MAY_READ | NFSD_MAY_READ_IF_EXEC)))
2082
		err = inode_permission(inode, MAY_EXEC);
Linus Torvalds's avatar
Linus Torvalds committed
2083 2084 2085 2086 2087 2088 2089

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2090 2091 2092
	struct raparms *raparm, *last_raparm;
	unsigned int i;

Linus Torvalds's avatar
Linus Torvalds committed
2093
	dprintk("nfsd: freeing readahead buffers.\n");
2094 2095 2096 2097 2098 2099 2100 2101 2102 2103

	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;
	}
Linus Torvalds's avatar
Linus Torvalds committed
2104 2105 2106 2107 2108 2109 2110 2111
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2112 2113
	int	j = 0;
	int	nperbucket;
2114
	struct raparms **raparm = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
2115

2116

2117
	if (raparm_hash[0].pb_head)
Linus Torvalds's avatar
Linus Torvalds committed
2118
		return 0;
2119
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
2120
	nperbucket = max(2, nperbucket);
2121
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2122 2123

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2124 2125

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2126
		spin_lock_init(&raparm_hash[i].pb_lock);
2127 2128 2129 2130 2131 2132 2133 2134 2135

		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;
2136 2137
	}

Linus Torvalds's avatar
Linus Torvalds committed
2138 2139
	nfsdstats.ra_size = cache_size;
	return 0;
2140 2141 2142 2143 2144

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
2145
}