vfs.c 52.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
#define MSNFS	/* HACK HACK */
/*
 * linux/fs/nfsd/vfs.c
 *
 * File operations used by nfsd. Some of these have been ripped from
 * other parts of the kernel because they weren't exported, others
 * are partial duplicates with added or changed functionality.
 *
 * Note that several functions dget() the dentry upon which they want
 * to act, most notably those that create directory entries. Response
 * dentry's are dput()'d if necessary in the release callback.
 * So if you notice code paths that apparently fail to dput() the
 * dentry, don't worry--they have been taken care of.
 *
 * Copyright (C) 1995-1999 Olaf Kirch <okir@monad.swb.de>
 * Zerocpy NFS support (C) 2002 Hirokazu Takahashi <taka@valinux.co.jp>
 */

#include <linux/string.h>
#include <linux/time.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/mount.h>
#include <linux/major.h>
26
#include <linux/splice.h>
L
Linus Torvalds 已提交
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
#include <linux/proc_fs.h>
#include <linux/stat.h>
#include <linux/fcntl.h>
#include <linux/net.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/in.h>
#include <linux/module.h>
#include <linux/namei.h>
#include <linux/vfs.h>
#include <linux/delay.h>
#include <linux/sunrpc/svc.h>
#include <linux/nfsd/nfsd.h>
#ifdef CONFIG_NFSD_V3
#include <linux/nfs3.h>
#include <linux/nfsd/xdr3.h>
#endif /* CONFIG_NFSD_V3 */
#include <linux/nfsd/nfsfh.h>
#include <linux/quotaops.h>
R
Robert Love 已提交
47
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
48 49 50
#include <linux/posix_acl.h>
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
51
#ifdef CONFIG_NFSD_V4
L
Linus Torvalds 已提交
52 53 54 55 56
#include <linux/nfs4.h>
#include <linux/nfs4_acl.h>
#include <linux/nfsd_idmap.h>
#include <linux/security.h>
#endif /* CONFIG_NFSD_V4 */
57
#include <linux/jhash.h>
L
Linus Torvalds 已提交
58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77

#include <asm/uaccess.h>

#define NFSDDBG_FACILITY		NFSDDBG_FILEOP


/*
 * This is a cache of readahead params that help us choose the proper
 * readahead strategy. Initially, we set all readahead parameters to 0
 * and let the VFS handle things.
 * If you increase the number of cached files very much, you'll need to
 * add a hash table here.
 */
struct raparms {
	struct raparms		*p_next;
	unsigned int		p_count;
	ino_t			p_ino;
	dev_t			p_dev;
	int			p_set;
	struct file_ra_state	p_ra;
78
	unsigned int		p_hindex;
L
Linus Torvalds 已提交
79 80
};

81 82 83 84 85 86 87 88 89
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
90 91 92 93

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
94
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102
 *  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;
103
	struct vfsmount *mnt = mntget(exp->ex_path.mnt);
L
Linus Torvalds 已提交
104
	struct dentry *mounts = dget(dentry);
105
	int err = 0;
L
Linus Torvalds 已提交
106 107 108

	while (follow_down(&mnt,&mounts)&&d_mountpoint(mounts));

109
	exp2 = rqst_exp_get_by_name(rqstp, mnt, mounts);
L
Linus Torvalds 已提交
110
	if (IS_ERR(exp2)) {
111 112
		if (PTR_ERR(exp2) != -ENOENT)
			err = PTR_ERR(exp2);
L
Linus Torvalds 已提交
113 114 115 116
		dput(mounts);
		mntput(mnt);
		goto out;
	}
117
	if ((exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
L
Linus Torvalds 已提交
118
		/* successfully crossed mount point */
A
Al Viro 已提交
119 120 121 122 123
		/*
		 * This is subtle: dentry is *not* under mnt at this point.
		 * The only reason we are safe is that original mnt is pinned
		 * down by exp, so we should dput before putting exp.
		 */
L
Linus Torvalds 已提交
124 125
		dput(dentry);
		*dpp = mounts;
A
Al Viro 已提交
126 127
		exp_put(exp);
		*expp = exp2;
L
Linus Torvalds 已提交
128
	} else {
129
		exp_put(exp2);
L
Linus Torvalds 已提交
130 131 132 133 134 135 136
		dput(mounts);
	}
	mntput(mnt);
out:
	return err;
}

137
__be32
138
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
139
		   const char *name, unsigned int len,
140
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
141 142 143 144
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
145 146
	__be32			err;
	int			host_err;
L
Linus Torvalds 已提交
147 148 149 150

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

	/* Obtain dentry and export. */
M
Miklos Szeredi 已提交
151
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
152 153 154 155 156 157 158 159 160 161 162
	if (err)
		return err;

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

	/* Lookup the name, but don't follow links */
	if (isdotent(name, len)) {
		if (len==1)
			dentry = dget(dparent);
163
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
164
			dentry = dget_parent(dparent);
165
		else if (!EX_NOHIDE(exp))
L
Linus Torvalds 已提交
166 167 168 169 170
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
			struct svc_export *exp2 = NULL;
			struct dentry *dp;
171
			struct vfsmount *mnt = mntget(exp->ex_path.mnt);
L
Linus Torvalds 已提交
172 173 174 175 176 177 178
			dentry = dget(dparent);
			while(dentry == mnt->mnt_root && follow_up(&mnt, &dentry))
				;
			dp = dget_parent(dentry);
			dput(dentry);
			dentry = dp;

179
			exp2 = rqst_exp_parent(rqstp, mnt, dentry);
180 181 182 183
			if (PTR_ERR(exp2) == -ENOENT) {
				dput(dentry);
				dentry = dget(dparent);
			} else if (IS_ERR(exp2)) {
184
				host_err = PTR_ERR(exp2);
L
Linus Torvalds 已提交
185 186 187 188 189 190 191 192 193 194 195 196
				dput(dentry);
				mntput(mnt);
				goto out_nfserr;
			} else {
				exp_put(exp);
				exp = exp2;
			}
			mntput(mnt);
		}
	} else {
		fh_lock(fhp);
		dentry = lookup_one_len(name, dparent, len);
197
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
198 199 200 201 202 203
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
		if (d_mountpoint(dentry)) {
204
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
205 206 207 208 209
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
	*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,
233
				unsigned int len, struct svc_fh *resfh)
234 235 236 237 238 239 240 241
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
242 243 244
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
245 246 247 248 249 250 251
	/*
	 * 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;
252
out:
L
Linus Torvalds 已提交
253 254 255 256 257
	dput(dentry);
	exp_put(exp);
	return err;
}

258

L
Linus Torvalds 已提交
259 260 261 262
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
263
__be32
L
Linus Torvalds 已提交
264 265 266 267 268
nfsd_setattr(struct svc_rqst *rqstp, struct svc_fh *fhp, struct iattr *iap,
	     int check_guard, time_t guardtime)
{
	struct dentry	*dentry;
	struct inode	*inode;
M
Miklos Szeredi 已提交
269
	int		accmode = NFSD_MAY_SATTR;
L
Linus Torvalds 已提交
270
	int		ftype = 0;
271 272
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
273 274 275
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
276
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
277 278 279 280 281
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
282
	if (err)
L
Linus Torvalds 已提交
283 284 285 286 287
		goto out;

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

288 289 290 291 292 293 294
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

295 296
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
297 298 299 300 301 302 303 304 305 306 307 308
	 * which only requires access, and "set-[ac]time-to-X" which
	 * requires ownership.
	 * So if it looks like it might be "set both to the same time which
	 * is close to now", and if inode_change_ok fails, then we
	 * convert to "set to now" instead of "set to explicit time"
	 *
	 * We only call inode_change_ok as the last test as technically
	 * it is not an interface that we should be using.  It is only
	 * valid if the filesystem does not define it's own i_op->setattr.
	 */
#define BOTH_TIME_SET (ATTR_ATIME_SET | ATTR_MTIME_SET)
#define	MAX_TOUCH_TIME_ERROR (30*60)
309 310 311 312 313 314 315 316
	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
317
		 */
318 319 320 321 322 323 324 325 326 327 328 329
		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
330 331
	}
	    
332 333 334 335
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
336 337
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
338 339
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
340 341 342 343 344 345 346 347
			if (err)
				goto out;
		}

		/*
		 * If we are changing the size of the file, then
		 * we need to break all leases.
		 */
348 349 350 351
		host_err = break_lease(inode, FMODE_WRITE | O_NONBLOCK);
		if (host_err == -EWOULDBLOCK)
			host_err = -ETIMEDOUT;
		if (host_err) /* ENOMEM or EWOULDBLOCK */
L
Linus Torvalds 已提交
352 353
			goto out_nfserr;

354 355
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
356 357 358
			goto out_nfserr;

		size_change = 1;
359 360
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
361 362 363
			put_write_access(inode);
			goto out_nfserr;
		}
364
		vfs_dq_init(inode);
L
Linus Torvalds 已提交
365 366
	}

367
	/* sanitize the mode change */
L
Linus Torvalds 已提交
368 369
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
370
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
371 372 373
	}

	/* Revoke setuid/setgid on chown */
374 375 376
	if (!S_ISDIR(inode->i_mode) &&
	    (((iap->ia_valid & ATTR_UID) && iap->ia_uid != inode->i_uid) ||
	     ((iap->ia_valid & ATTR_GID) && iap->ia_gid != inode->i_gid))) {
377 378 379
		iap->ia_valid |= ATTR_KILL_PRIV;
		if (iap->ia_valid & ATTR_MODE) {
			/* we're setting mode too, just clear the s*id bits */
380
			iap->ia_mode &= ~S_ISUID;
381 382 383 384 385
			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);
386
		}
L
Linus Torvalds 已提交
387 388 389 390 391 392 393 394 395
	}

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
		fh_lock(fhp);
396 397
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
398 399 400 401 402 403 404 405 406 407 408
		fh_unlock(fhp);
	}
	if (size_change)
		put_write_access(inode);
	if (!err)
		if (EX_ISSYNC(fhp->fh_export))
			write_inode_now(inode, 1);
out:
	return err;

out_nfserr:
409
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
410 411 412
	goto out;
}

413 414 415 416 417 418
#if defined(CONFIG_NFSD_V2_ACL) || \
    defined(CONFIG_NFSD_V3_ACL) || \
    defined(CONFIG_NFSD_V4)
static ssize_t nfsd_getxattr(struct dentry *dentry, char *key, void **buf)
{
	ssize_t buflen;
419
	ssize_t ret;
420 421 422 423

	buflen = vfs_getxattr(dentry, key, NULL, 0);
	if (buflen <= 0)
		return buflen;
L
Linus Torvalds 已提交
424

425 426 427 428
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

429 430 431 432
	ret = vfs_getxattr(dentry, key, *buf, buflen);
	if (ret < 0)
		kfree(*buf);
	return ret;
433 434 435 436
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
static int
set_nfsv4_acl_one(struct dentry *dentry, struct posix_acl *pacl, char *key)
{
	int len;
	size_t buflen;
	char *buf = NULL;
	int error = 0;

	buflen = posix_acl_xattr_size(pacl->a_count);
	buf = kmalloc(buflen, GFP_KERNEL);
	error = -ENOMEM;
	if (buf == NULL)
		goto out;

	len = posix_acl_to_xattr(pacl, buf, buflen);
	if (len < 0) {
		error = len;
		goto out;
	}

457
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
458 459 460 461 462
out:
	kfree(buf);
	return error;
}

463
__be32
L
Linus Torvalds 已提交
464 465 466
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
467 468
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
469 470 471 472 473 474
	struct dentry *dentry;
	struct inode *inode;
	struct posix_acl *pacl = NULL, *dpacl = NULL;
	unsigned int flags = 0;

	/* Get inode */
M
Miklos Szeredi 已提交
475
	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
476
	if (error)
477
		return error;
L
Linus Torvalds 已提交
478 479 480 481 482 483

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

484 485
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
486
		return nfserr_attrnotsupp;
487
	} else if (host_error < 0)
L
Linus Torvalds 已提交
488 489
		goto out_nfserr;

490 491
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
492
		goto out_release;
L
Linus Torvalds 已提交
493

494
	if (S_ISDIR(inode->i_mode))
495
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
496

497
out_release:
L
Linus Torvalds 已提交
498 499 500
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
501
	if (host_error == -EOPNOTSUPP)
502
		return nfserr_attrnotsupp;
503
	else
504
		return nfserrno(host_error);
L
Linus Torvalds 已提交
505 506 507 508 509
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
510
	void *buf = NULL;
L
Linus Torvalds 已提交
511
	struct posix_acl *pacl = NULL;
512
	int buflen;
L
Linus Torvalds 已提交
513

514 515 516 517 518
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
519 520 521 522 523 524 525 526 527 528 529 530 531 532

	pacl = posix_acl_from_xattr(buf, buflen);
	kfree(buf);
	return pacl;
}

int
nfsd4_get_nfs4_acl(struct svc_rqst *rqstp, struct dentry *dentry, struct nfs4_acl **acl)
{
	struct inode *inode = dentry->d_inode;
	int error = 0;
	struct posix_acl *pacl = NULL, *dpacl = NULL;
	unsigned int flags = 0;

533
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
534 535 536 537 538 539 540 541 542
	if (IS_ERR(pacl) && PTR_ERR(pacl) == -ENODATA)
		pacl = posix_acl_from_mode(inode->i_mode, GFP_KERNEL);
	if (IS_ERR(pacl)) {
		error = PTR_ERR(pacl);
		pacl = NULL;
		goto out;
	}

	if (S_ISDIR(inode->i_mode)) {
543
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575
		if (IS_ERR(dpacl) && PTR_ERR(dpacl) == -ENODATA)
			dpacl = NULL;
		else if (IS_ERR(dpacl)) {
			error = PTR_ERR(dpacl);
			dpacl = NULL;
			goto out;
		}
		flags = NFS4_ACL_DIR;
	}

	*acl = nfs4_acl_posix_to_nfsv4(pacl, dpacl, flags);
	if (IS_ERR(*acl)) {
		error = PTR_ERR(*acl);
		*acl = NULL;
	}
 out:
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
	return error;
}

#endif /* defined(CONFIG_NFS_V4) */

#ifdef CONFIG_NFSD_V3
/*
 * Check server access rights to a file system object
 */
struct accessmap {
	u32		access;
	int		how;
};
static struct accessmap	nfs3_regaccess[] = {
M
Miklos Szeredi 已提交
576 577 578 579
    {	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
580 581 582 583 584

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
585 586 587 588 589
    {	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
590 591 592 593 594 595 596 597 598 599 600 601

    {	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
602 603 604 605
    {	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
606 607 608 609

    {	0,			0				}
};

610
__be32
L
Linus Torvalds 已提交
611 612 613 614 615 616
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;
617
	__be32			error;
L
Linus Torvalds 已提交
618

M
Miklos Szeredi 已提交
619
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
	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) {
637
			__be32 err2;
L
Linus Torvalds 已提交
638 639 640

			sresult |= map->access;

641
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
			switch (err2) {
			case nfs_ok:
				result |= map->access;
				break;
				
			/* the following error codes just mean the access was not allowed,
			 * rather than an error occurred */
			case nfserr_rofs:
			case nfserr_acces:
			case nfserr_perm:
				/* simply don't "or" in the access bit. */
				break;
			default:
				error = err2;
				goto out;
			}
		}
	}
	*access = result;
	if (supported)
		*supported = sresult;

 out:
	return error;
}
#endif /* CONFIG_NFSD_V3 */



/*
 * Open an existing file or directory.
 * The access argument indicates the type of open (read/write/lock)
 * N.B. After this call fhp needs an fh_put
 */
676
__be32
L
Linus Torvalds 已提交
677 678 679
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
			int access, struct file **filp)
{
680
	const struct cred *cred = current_cred();
L
Linus Torvalds 已提交
681 682
	struct dentry	*dentry;
	struct inode	*inode;
683 684 685
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
686 687 688 689 690 691

	/*
	 * If we get here, then the client has already done an "open",
	 * and (hopefully) checked permission - so allow OWNER_OVERRIDE
	 * in case a chmod has now revoked permission.
	 */
M
Miklos Szeredi 已提交
692
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
693 694 695 696 697 698 699 700 701 702
	if (err)
		goto out;

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

	/* Disallow write access to files with the append-only bit set
	 * or any access when mandatory locking enabled
	 */
	err = nfserr_perm;
M
Miklos Szeredi 已提交
703
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
704
		goto out;
J
J. Bruce Fields 已提交
705 706 707 708 709 710
	/*
	 * 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
711 712 713 714 715 716 717 718 719
		goto out;

	if (!inode->i_fop)
		goto out;

	/*
	 * Check to see if there are any leases on this file.
	 * This may block while leases are broken.
	 */
M
Miklos Szeredi 已提交
720
	host_err = break_lease(inode, O_NONBLOCK | ((access & NFSD_MAY_WRITE) ? FMODE_WRITE : 0));
721 722 723
	if (host_err == -EWOULDBLOCK)
		host_err = -ETIMEDOUT;
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
724 725
		goto out_nfserr;

M
Miklos Szeredi 已提交
726 727
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
728 729 730
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
731

732
		vfs_dq_init(inode);
L
Linus Torvalds 已提交
733
	}
734
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
735
			    flags, cred);
L
Linus Torvalds 已提交
736
	if (IS_ERR(*filp))
737
		host_err = PTR_ERR(*filp);
L
Linus Torvalds 已提交
738
out_nfserr:
739
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
740 741 742 743 744 745 746 747 748 749 750 751 752
out:
	return err;
}

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

753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773
/*
 * Sync a file
 * As this calls fsync (not fdatasync) there is no need for a write_inode
 * after it.
 */
static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
			      const struct file_operations *fop)
{
	struct inode *inode = dp->d_inode;
	int (*fsync) (struct file *, struct dentry *, int);
	int err;

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

	return err;
}

774
static int
L
Linus Torvalds 已提交
775 776
nfsd_sync(struct file *filp)
{
777 778 779 780 781 782 783 784
        int err;
	struct inode *inode = filp->f_path.dentry->d_inode;
	dprintk("nfsd: sync file %s\n", filp->f_path.dentry->d_name.name);
	mutex_lock(&inode->i_mutex);
	err=nfsd_dosync(filp, filp->f_path.dentry, filp->f_op);
	mutex_unlock(&inode->i_mutex);

	return err;
L
Linus Torvalds 已提交
785 786
}

787
int
788
nfsd_sync_dir(struct dentry *dp)
L
Linus Torvalds 已提交
789
{
790
	return nfsd_dosync(NULL, dp, dp->d_inode->i_fop);
L
Linus Torvalds 已提交
791 792 793 794 795 796 797 798 799 800 801 802
}

/*
 * 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;
803 804 805 806 807
	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
808

809 810
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
811 812 813 814 815 816 817 818
		if (ra->p_ino == ino && ra->p_dev == dev)
			goto found;
		depth++;
		if (ra->p_count == 0)
			frap = rap;
	}
	depth = nfsdstats.ra_size*11/10;
	if (!frap) {	
819
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
820 821 822 823 824 825 826
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
827
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
828
found:
829
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
830
		*rap = ra->p_next;
831 832
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
833 834 835
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
836
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
837 838 839 840
	return ra;
}

/*
841 842 843
 * 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
844 845
 */
static int
846 847
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
848
{
849
	struct svc_rqst *rqstp = sd->u.data;
850
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
851 852 853
	struct page *page = buf->page;
	size_t size;
	int ret;
L
Linus Torvalds 已提交
854

855
	ret = buf->ops->confirm(pipe, buf);
856 857 858 859
	if (unlikely(ret))
		return ret;

	size = sd->len;
L
Linus Torvalds 已提交
860 861 862

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
863 864 865
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
866
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
867
		rqstp->rq_res.page_len = size;
868
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
869
		get_page(page);
870 871
		if (*pp)
			put_page(*pp);
872 873
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
874
		rqstp->rq_res.page_len += size;
875
	} else
L
Linus Torvalds 已提交
876 877 878 879 880
		rqstp->rq_res.page_len += size;

	return size;
}

881 882 883 884 885 886
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

887 888 889 890 891 892 893 894 895
static inline int svc_msnfs(struct svc_fh *ffhp)
{
#ifdef MSNFS
	return (ffhp->fh_export->ex_flags & NFSEXP_MSNFS);
#else
	return 0;
#endif
}

896
static __be32
L
Linus Torvalds 已提交
897 898 899 900 901 902
nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
              loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
{
	struct inode *inode;
	struct raparms	*ra;
	mm_segment_t	oldfs;
903 904
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
905 906

	err = nfserr_perm;
907
	inode = file->f_path.dentry->d_inode;
908 909

	if (svc_msnfs(fhp) && !lock_may_read(inode, offset, *count))
L
Linus Torvalds 已提交
910 911 912 913 914 915 916 917
		goto out;

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

	if (ra && ra->p_set)
		file->f_ra = ra->p_ra;

918 919 920 921 922 923 924 925
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

926
		rqstp->rq_resused = 1;
927
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
928 929 930
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
931
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
932 933 934 935 936
		set_fs(oldfs);
	}

	/* Write back readahead params */
	if (ra) {
937 938
		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
		spin_lock(&rab->pb_lock);
L
Linus Torvalds 已提交
939 940 941
		ra->p_ra = file->f_ra;
		ra->p_set = 1;
		ra->p_count--;
942
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
943 944
	}

945 946 947
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
948
		err = 0;
949
		fsnotify_access(file->f_path.dentry);
L
Linus Torvalds 已提交
950
	} else 
951
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
952 953 954 955
out:
	return err;
}

956 957 958
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
959
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
960

961
	mutex_lock(&dentry->d_inode->i_mutex);
962
	notify_change(dentry, &ia);
963
	mutex_unlock(&dentry->d_inode->i_mutex);
964 965
}

966
static __be32
L
Linus Torvalds 已提交
967 968
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
969
				unsigned long *cnt, int *stablep)
L
Linus Torvalds 已提交
970 971 972 973 974
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
975 976
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
977 978
	int			stable = *stablep;

979
#ifdef MSNFS
L
Linus Torvalds 已提交
980 981 982
	err = nfserr_perm;

	if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
983
		(!lock_may_write(file->f_path.dentry->d_inode, offset, *cnt)))
L
Linus Torvalds 已提交
984 985 986
		goto out;
#endif

987
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
988 989 990 991 992 993 994 995 996 997 998 999
	inode = dentry->d_inode;
	exp   = fhp->fh_export;

	/*
	 * Request sync writes if
	 *  -	the sync export option has been set, or
	 *  -	the client requested O_SYNC behavior (NFSv3 feature).
	 *  -   The file system doesn't support fsync().
	 * When gathered writes have been configured for this volume,
	 * flushing the data to disk is handled separately below.
	 */

1000
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
1001 1002 1003 1004 1005 1006
	       stable = 2;
	       *stablep = 2; /* FILE_SYNC */
	}

	if (!EX_ISSYNC(exp))
		stable = 0;
J
Jonathan Corbet 已提交
1007 1008
	if (stable && !EX_WGATHER(exp)) {
		spin_lock(&file->f_lock);
L
Linus Torvalds 已提交
1009
		file->f_flags |= O_SYNC;
J
Jonathan Corbet 已提交
1010 1011
		spin_unlock(&file->f_lock);
	}
L
Linus Torvalds 已提交
1012 1013 1014

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
1015
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1016
	set_fs(oldfs);
1017
	if (host_err >= 0) {
1018
		*cnt = host_err;
1019
		nfsdstats.io_write += host_err;
1020
		fsnotify_modify(file->f_path.dentry);
L
Linus Torvalds 已提交
1021 1022 1023
	}

	/* clear setuid/setgid flag after write */
1024
	if (host_err >= 0 && (inode->i_mode & (S_ISUID | S_ISGID)))
1025
		kill_suid(dentry);
L
Linus Torvalds 已提交
1026

1027
	if (host_err >= 0 && stable) {
L
Linus Torvalds 已提交
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
		static ino_t	last_ino;
		static dev_t	last_dev;

		/*
		 * Gathered writes: If another process is currently
		 * writing to the file, there's a high chance
		 * this is another nfsd (triggered by a bulk write
		 * from a client's biod). Rather than syncing the
		 * file with each write request, we sleep for 10 msec.
		 *
		 * I don't know if this roughly approximates
		 * C. Juszak's idea of gathered writes, but it's a
		 * nice and simple solution (IMHO), and it seems to
		 * work:-)
		 */
		if (EX_WGATHER(exp)) {
			if (atomic_read(&inode->i_writecount) > 1
			    || (last_ino == inode->i_ino && last_dev == inode->i_sb->s_dev)) {
1046
				dprintk("nfsd: write defer %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1047
				msleep(10);
1048
				dprintk("nfsd: write resume %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1049 1050 1051
			}

			if (inode->i_state & I_DIRTY) {
1052
				dprintk("nfsd: write sync %d\n", task_pid_nr(current));
1053
				host_err=nfsd_sync(file);
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062
			}
#if 0
			wake_up(&inode->i_wait);
#endif
		}
		last_ino = inode->i_ino;
		last_dev = inode->i_sb->s_dev;
	}

1063
	dprintk("nfsd: write complete host_err=%d\n", host_err);
1064
	if (host_err >= 0)
L
Linus Torvalds 已提交
1065
		err = 0;
1066
	else
1067
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1068 1069 1070 1071 1072 1073 1074 1075 1076
out:
	return err;
}

/*
 * Read data from a file. count must contain the requested read count
 * on entry. On return, *count contains the number of bytes actually read.
 * N.B. After this call fhp needs an fh_put
 */
1077
__be32
L
Linus Torvalds 已提交
1078 1079 1080 1081
nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1082
	__be32		err;
L
Linus Torvalds 已提交
1083 1084

	if (file) {
1085
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1086
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1087 1088 1089 1090
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
	} else {
M
Miklos Szeredi 已提交
1091
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
		nfsd_close(file);
	}
out:
	return err;
}

/*
 * Write data to a file.
 * The stable flag requests synchronous writes.
 * N.B. After this call fhp needs an fh_put
 */
1106
__be32
L
Linus Torvalds 已提交
1107
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1108
		loff_t offset, struct kvec *vec, int vlen, unsigned long *cnt,
L
Linus Torvalds 已提交
1109 1110
		int *stablep)
{
1111
	__be32			err = 0;
L
Linus Torvalds 已提交
1112 1113

	if (file) {
1114
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1115
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1116 1117 1118 1119 1120
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1121
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142
		if (err)
			goto out;

		if (cnt)
			err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen,
					     cnt, stablep);
		nfsd_close(file);
	}
out:
	return err;
}

#ifdef CONFIG_NFSD_V3
/*
 * Commit all pending writes to stable storage.
 * Strictly speaking, we could sync just the indicated file region here,
 * but there's currently no way we can ask the VFS to do so.
 *
 * Unfortunately we cannot lock the file to make sure we return full WCC
 * data to the client, as locking happens lower down in the filesystem.
 */
1143
__be32
L
Linus Torvalds 已提交
1144 1145 1146 1147
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1148
	__be32		err;
L
Linus Torvalds 已提交
1149 1150 1151 1152

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

M
Miklos Szeredi 已提交
1153 1154
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
	if (err)
L
Linus Torvalds 已提交
1155 1156 1157
		return err;
	if (EX_ISSYNC(fhp->fh_export)) {
		if (file->f_op && file->f_op->fsync) {
1158
			err = nfserrno(nfsd_sync(file));
L
Linus Torvalds 已提交
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168
		} else {
			err = nfserr_notsupp;
		}
	}

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

A
Adrian Bunk 已提交
1169
static __be32
1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
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:
	 */
1182
	if (current_fsuid() != 0)
1183 1184 1185 1186 1187 1188
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203
/* 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
1204 1205 1206 1207 1208 1209 1210 1211
/*
 * 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
 */
1212
__be32
L
Linus Torvalds 已提交
1213 1214 1215 1216 1217 1218
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;
1219
	__be32		err;
1220
	__be32		err2;
1221
	int		host_err;
L
Linus Torvalds 已提交
1222 1223 1224 1225 1226 1227 1228 1229

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

M
Miklos Szeredi 已提交
1230
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1231 1232 1233 1234 1235 1236 1237
	if (err)
		goto out;

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

	err = nfserr_notdir;
A
Al Viro 已提交
1238
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1239 1240 1241 1242 1243 1244 1245
		goto out;
	/*
	 * Check whether the response file handle has been verified yet.
	 * If it has, the parent directory should already be locked.
	 */
	if (!resfhp->fh_dentry) {
		/* called from nfsd_proc_mkdir, or possibly nfsd3_proc_create */
P
Peter Zijlstra 已提交
1246
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1247
		dchild = lookup_one_len(fname, dentry, flen);
1248
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
		if (IS_ERR(dchild))
			goto out_nfserr;
		err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
		if (err)
			goto out;
	} else {
		/* called from nfsd_proc_create */
		dchild = dget(resfhp->fh_dentry);
		if (!fhp->fh_locked) {
			/* not actually possible */
			printk(KERN_ERR
				"nfsd_create: parent %s/%s not locked!\n",
				dentry->d_parent->d_name.name,
				dentry->d_name.name);
A
Al Viro 已提交
1263
			err = nfserr_io;
L
Linus Torvalds 已提交
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
			goto out;
		}
	}
	/*
	 * Make sure the child dentry is still negative ...
	 */
	err = nfserr_exist;
	if (dchild->d_inode) {
		dprintk("nfsd_create: dentry %s/%s not negative!\n",
			dentry->d_name.name, dchild->d_name.name);
		goto out; 
	}

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

1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
	err = nfserr_inval;
	if (!S_ISREG(type) && !S_ISDIR(type) && !special_file(type)) {
		printk(KERN_WARNING "nfsd: bad file type %o in nfsd_create\n",
		       type);
		goto out;
	}

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

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

	if (EX_ISSYNC(fhp->fh_export)) {
1318
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1319 1320 1321
		write_inode_now(dchild->d_inode, 1);
	}

1322 1323 1324
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1325
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
	/*
	 * 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:
1337
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1338 1339 1340 1341 1342 1343 1344
	goto out;
}

#ifdef CONFIG_NFSD_V3
/*
 * NFSv3 version of nfsd_create
 */
1345
__be32
L
Linus Torvalds 已提交
1346 1347 1348
nfsd_create_v3(struct svc_rqst *rqstp, struct svc_fh *fhp,
		char *fname, int flen, struct iattr *iap,
		struct svc_fh *resfhp, int createmode, u32 *verifier,
1349
	        int *truncp, int *created)
L
Linus Torvalds 已提交
1350 1351 1352
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1353
	__be32		err;
1354
	__be32		err2;
1355
	int		host_err;
L
Linus Torvalds 已提交
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
	__u32		v_mtime=0, v_atime=0;

	err = nfserr_perm;
	if (!flen)
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;
	if (!(iap->ia_valid & ATTR_MODE))
		iap->ia_mode = 0;
M
Miklos Szeredi 已提交
1366
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375
	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
1376
	if (!dirp->i_op->lookup)
L
Linus Torvalds 已提交
1377
		goto out;
P
Peter Zijlstra 已提交
1378
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1379 1380 1381 1382 1383

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1384
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1385 1386 1387 1388 1389 1390 1391 1392
	if (IS_ERR(dchild))
		goto out_nfserr;

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

	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1393
		/* solaris7 gets confused (bugid 4218508) if these have
1394 1395 1396 1397
		 * 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
1398 1399 1400 1401 1402
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1403 1404 1405
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
	if (dchild->d_inode) {
		err = 0;

		switch (createmode) {
		case NFS3_CREATE_UNCHECKED:
			if (! S_ISREG(dchild->d_inode->i_mode))
				err = nfserr_exist;
			else if (truncp) {
				/* in nfsv4, we need to treat this case a little
				 * differently.  we don't want to truncate the
				 * file now; this would be wrong if the OPEN
				 * fails for some other reason.  furthermore,
				 * if the size is nonzero, we should ignore it
				 * according to spec!
				 */
				*truncp = (iap->ia_valid & ATTR_SIZE) && !iap->ia_size;
			}
			else {
				iap->ia_valid &= ATTR_SIZE;
				goto set_attr;
			}
			break;
		case NFS3_CREATE_EXCLUSIVE:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
			    && dchild->d_inode->i_size  == 0 )
				break;
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
1437
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1438 1439 1440
		goto out;
	}

1441
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1442 1443
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1444
		goto out_nfserr;
1445
	}
1446 1447
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1448 1449

	if (EX_ISSYNC(fhp->fh_export)) {
1450
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1451 1452 1453
		/* setattr will sync the child (or not) */
	}

1454 1455
	nfsd_check_ignore_resizing(iap);

L
Linus Torvalds 已提交
1456
	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1457
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1458
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1459
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1460 1461 1462 1463 1464 1465 1466 1467
		/* 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:
1468 1469 1470
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1471

1472
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
1473 1474 1475 1476 1477
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1478 1479 1480 1481 1482 1483 1484 1485

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1486
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1487 1488 1489 1490 1491 1492 1493 1494 1495
	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
 */
1496
__be32
L
Linus Torvalds 已提交
1497 1498 1499 1500 1501
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1502 1503
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1504

M
Miklos Szeredi 已提交
1505
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1506 1507 1508 1509 1510 1511 1512
	if (err)
		goto out;

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

	err = nfserr_inval;
A
Al Viro 已提交
1513
	if (!inode->i_op->readlink)
L
Linus Torvalds 已提交
1514 1515
		goto out;

1516
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1517 1518 1519 1520 1521
	/* 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);
1522
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1523 1524
	set_fs(oldfs);

1525
	if (host_err < 0)
L
Linus Torvalds 已提交
1526
		goto out_nfserr;
1527
	*lenp = host_err;
L
Linus Torvalds 已提交
1528 1529 1530 1531 1532
	err = 0;
out:
	return err;

out_nfserr:
1533
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1534 1535 1536 1537 1538 1539 1540
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1541
__be32
L
Linus Torvalds 已提交
1542 1543 1544 1545 1546 1547 1548
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;
1549 1550
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1551 1552 1553 1554 1555 1556 1557 1558

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

M
Miklos Szeredi 已提交
1559
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1560 1561 1562 1563 1564
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1565
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1566 1567 1568
	if (IS_ERR(dnew))
		goto out_nfserr;

1569 1570 1571 1572
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1573 1574 1575
	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1576
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1577 1578 1579
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1580
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1581 1582 1583
			kfree(path_alloced);
		}
	} else
1584
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
L
Linus Torvalds 已提交
1585

1586
	if (!host_err) {
L
Linus Torvalds 已提交
1587
		if (EX_ISSYNC(fhp->fh_export))
1588 1589 1590
			host_err = nfsd_sync_dir(dentry);
	}
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1591 1592
	fh_unlock(fhp);

1593 1594
	mnt_drop_write(fhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1595 1596 1597 1598 1599 1600 1601
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1602
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1603 1604 1605 1606 1607 1608 1609
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1610
__be32
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
	struct inode	*dirp, *dest;
1616 1617
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1618

M
Miklos Szeredi 已提交
1619
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1620 1621
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1622
	err = fh_verify(rqstp, tfhp, -S_IFDIR, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
	if (err)
		goto out;

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

P
Peter Zijlstra 已提交
1633
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1634 1635 1636 1637
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1638
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1639 1640 1641 1642 1643 1644
	if (IS_ERR(dnew))
		goto out_nfserr;

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

1645 1646 1647 1648 1649
	host_err = mnt_want_write(tfhp->fh_export->ex_path.mnt);
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1650 1651
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
L
Linus Torvalds 已提交
1652
		if (EX_ISSYNC(ffhp->fh_export)) {
1653
			err = nfserrno(nfsd_sync_dir(ddir));
L
Linus Torvalds 已提交
1654 1655
			write_inode_now(dest, 1);
		}
1656
		err = 0;
L
Linus Torvalds 已提交
1657
	} else {
1658
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1659 1660
			err = nfserr_acces;
		else
1661
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1662
	}
1663 1664
	mnt_drop_write(tfhp->fh_export->ex_path.mnt);
out_dput:
L
Linus Torvalds 已提交
1665
	dput(dnew);
1666 1667
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1668 1669 1670 1671
out:
	return err;

out_nfserr:
1672
	err = nfserrno(host_err);
1673
	goto out_unlock;
L
Linus Torvalds 已提交
1674 1675 1676 1677 1678 1679
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1680
__be32
L
Linus Torvalds 已提交
1681 1682 1683 1684 1685
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;
1686 1687
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1688

M
Miklos Szeredi 已提交
1689
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1690 1691
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1692
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
	if (err)
		goto out;

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

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

	err = (rqstp->rq_vers == 2) ? nfserr_acces : nfserr_xdev;
1703
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
		goto out;

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

	/* cannot use fh_lock as we need deadlock protective ordering
	 * so do it by hand */
	trap = lock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 1;
	fill_pre_wcc(ffhp);
	fill_pre_wcc(tfhp);

	odentry = lookup_one_len(fname, fdentry, flen);
1718
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1719 1720 1721
	if (IS_ERR(odentry))
		goto out_nfserr;

1722
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1723 1724
	if (!odentry->d_inode)
		goto out_dput_old;
1725
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1726 1727 1728 1729
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1730
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1731 1732
	if (IS_ERR(ndentry))
		goto out_dput_old;
1733
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1734 1735 1736
	if (ndentry == trap)
		goto out_dput_new;

1737
	if (svc_msnfs(ffhp) &&
L
Linus Torvalds 已提交
1738 1739
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
1740
			host_err = -EPERM;
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750
			goto out_dput_new;
	}

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

1751 1752 1753 1754 1755
	host_err = vfs_rename(fdir, odentry, tdir, ndentry);
	if (!host_err && EX_ISSYNC(tfhp->fh_export)) {
		host_err = nfsd_sync_dir(tdentry);
		if (!host_err)
			host_err = nfsd_sync_dir(fdentry);
L
Linus Torvalds 已提交
1756 1757
	}

1758 1759
	mnt_drop_write(ffhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1760 1761 1762 1763 1764
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1765
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783

	/* we cannot reply on fh_unlock on the two filehandles,
	 * as that would do the wrong thing if the two directories
	 * were the same, so again we do it by hand
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
1784
__be32
L
Linus Torvalds 已提交
1785 1786 1787 1788 1789
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1790 1791
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1792 1793 1794 1795

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1796
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1797 1798 1799
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1800
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1801 1802 1803 1804
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1805
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817
	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;

1818 1819 1820 1821
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1822 1823 1824 1825
	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
1826
			host_err = -EPERM;
L
Linus Torvalds 已提交
1827 1828
		} else
#endif
1829
		host_err = vfs_unlink(dirp, rdentry);
L
Linus Torvalds 已提交
1830
	} else { /* It's RMDIR */
1831
		host_err = vfs_rmdir(dirp, rdentry);
L
Linus Torvalds 已提交
1832 1833 1834 1835
	}

	dput(rdentry);

1836
	if (host_err)
1837
		goto out_drop;
1838 1839
	if (EX_ISSYNC(fhp->fh_export))
		host_err = nfsd_sync_dir(dentry);
L
Linus Torvalds 已提交
1840

1841 1842
out_drop:
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1843
out_nfserr:
1844
	err = nfserrno(host_err);
1845 1846
out:
	return err;
L
Linus Torvalds 已提交
1847 1848
}

1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866
/*
 * We do this buffering because we must not call back into the file
 * system's ->lookup() method from the filldir callback. That may well
 * deadlock a number of file systems.
 *
 * This is based heavily on the implementation of same in XFS.
 */
struct buffered_dirent {
	u64		ino;
	loff_t		offset;
	int		namlen;
	unsigned int	d_type;
	char		name[];
};

struct readdir_data {
	char		*dirent;
	size_t		used;
1867
	int		full;
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
};

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));
1878 1879
	if (buf->used + reclen > PAGE_SIZE) {
		buf->full = 1;
1880
		return -EINVAL;
1881
	}
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892

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

	return 0;
}

1893 1894
static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func,
				    struct readdir_cd *cdp, loff_t *offsetp)
1895
{
1896 1897
	struct readdir_data buf;
	struct buffered_dirent *de;
1898
	int host_err;
1899 1900
	int size;
	loff_t offset;
1901

1902 1903
	buf.dirent = (void *)__get_free_page(GFP_KERNEL);
	if (!buf.dirent)
1904
		return nfserrno(-ENOMEM);
1905 1906

	offset = *offsetp;
1907

1908
	while (1) {
1909
		struct inode *dir_inode = file->f_path.dentry->d_inode;
1910 1911
		unsigned int reclen;

1912
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1913
		buf.used = 0;
1914
		buf.full = 0;
1915 1916

		host_err = vfs_readdir(file, nfsd_buffered_filldir, &buf);
1917 1918 1919 1920
		if (buf.full)
			host_err = 0;

		if (host_err < 0)
1921 1922 1923 1924 1925 1926 1927
			break;

		size = buf.used;

		if (!size)
			break;

1928 1929 1930 1931 1932 1933 1934 1935 1936
		/*
		 * 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;

1937 1938 1939 1940 1941 1942
		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))
1943
				break;
1944 1945

			if (cdp->err != nfs_ok)
1946
				break;
1947 1948 1949 1950 1951 1952

			reclen = ALIGN(sizeof(*de) + de->namlen,
				       sizeof(u64));
			size -= reclen;
			de = (struct buffered_dirent *)((char *)de + reclen);
		}
1953 1954 1955 1956
		mutex_unlock(&dir_inode->i_mutex);
		if (size > 0) /* We bailed out early */
			break;

1957
		offset = vfs_llseek(file, 0, SEEK_CUR);
1958 1959 1960
	}

	free_page((unsigned long)(buf.dirent));
1961 1962 1963

	if (host_err)
		return nfserrno(host_err);
1964 1965 1966

	*offsetp = offset;
	return cdp->err;
1967 1968
}

L
Linus Torvalds 已提交
1969 1970 1971 1972
/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1973
__be32
L
Linus Torvalds 已提交
1974
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1975
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1976
{
1977
	__be32		err;
L
Linus Torvalds 已提交
1978 1979 1980
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
1981
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1982 1983 1984 1985 1986 1987 1988 1989 1990
	if (err)
		goto out;

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

1991
	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
L
Linus Torvalds 已提交
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

	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
 */
2005
__be32
2006
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
2007
{
2008
	__be32 err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
2009
	if (!err && vfs_statfs(fhp->fh_dentry,stat))
L
Linus Torvalds 已提交
2010 2011 2012 2013
		err = nfserr_io;
	return err;
}

J
J. Bruce Fields 已提交
2014
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
2015
{
J
J. Bruce Fields 已提交
2016
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
2017 2018
}

L
Linus Torvalds 已提交
2019 2020 2021
/*
 * Check for a user's access permissions to this inode.
 */
2022
__be32
2023 2024
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
2025 2026 2027 2028
{
	struct inode	*inode = dentry->d_inode;
	int		err;

M
Miklos Szeredi 已提交
2029
	if (acc == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
2030 2031 2032 2033
		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
2034 2035 2036 2037 2038 2039 2040
		(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
2041 2042 2043
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
2044
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
2045
	dprintk("      owner %d/%d user %d/%d\n",
2046
		inode->i_uid, inode->i_gid, current_fsuid(), current_fsgid());
L
Linus Torvalds 已提交
2047 2048 2049 2050 2051 2052
#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
2053 2054
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
2055 2056
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
2057
				return nfserr_rofs;
M
Miklos Szeredi 已提交
2058
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
2059 2060
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
2061
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
2062 2063
		return nfserr_perm;

M
Miklos Szeredi 已提交
2064
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
2065 2066 2067 2068 2069 2070 2071
		/* 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
2072
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
	}
	/*
	 * 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
2088
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
2089
	    inode->i_uid == current_fsuid())
L
Linus Torvalds 已提交
2090 2091
		return 0;

M
Miklos Szeredi 已提交
2092
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
2093
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
2094 2095 2096

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
M
Miklos Szeredi 已提交
2097
	    acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE))
2098
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
2099 2100 2101 2102 2103 2104 2105

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
2106 2107 2108
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
2109
	dprintk("nfsd: freeing readahead buffers.\n");
2110 2111 2112 2113 2114 2115 2116 2117 2118 2119

	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
2120 2121 2122 2123 2124 2125 2126 2127
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
2128 2129
	int	j = 0;
	int	nperbucket;
2130
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
2131

2132

2133
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
2134
		return 0;
2135 2136 2137 2138
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2139 2140

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2141 2142

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2143
		spin_lock_init(&raparm_hash[i].pb_lock);
2144 2145 2146 2147 2148 2149 2150 2151 2152

		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;
2153 2154
	}

L
Linus Torvalds 已提交
2155 2156
	nfsdstats.ra_size = cache_size;
	return 0;
2157 2158 2159 2160 2161

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2162
}
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173

#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
struct posix_acl *
nfsd_get_posix_acl(struct svc_fh *fhp, int type)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	char *name;
	void *value = NULL;
	ssize_t size;
	struct posix_acl *acl;

2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184
	if (!IS_POSIXACL(inode))
		return ERR_PTR(-EOPNOTSUPP);

	switch (type) {
	case ACL_TYPE_ACCESS:
		name = POSIX_ACL_XATTR_ACCESS;
		break;
	case ACL_TYPE_DEFAULT:
		name = POSIX_ACL_XATTR_DEFAULT;
		break;
	default:
2185 2186 2187
		return ERR_PTR(-EOPNOTSUPP);
	}

2188 2189 2190
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205

	acl = posix_acl_from_xattr(value, size);
	kfree(value);
	return acl;
}

int
nfsd_set_posix_acl(struct svc_fh *fhp, int type, struct posix_acl *acl)
{
	struct inode *inode = fhp->fh_dentry->d_inode;
	char *name;
	void *value = NULL;
	size_t size;
	int error;

A
Al Viro 已提交
2206
	if (!IS_POSIXACL(inode) ||
2207 2208 2209 2210
	    !inode->i_op->setxattr || !inode->i_op->removexattr)
		return -EOPNOTSUPP;
	switch(type) {
		case ACL_TYPE_ACCESS:
2211
			name = POSIX_ACL_XATTR_ACCESS;
2212 2213
			break;
		case ACL_TYPE_DEFAULT:
2214
			name = POSIX_ACL_XATTR_DEFAULT;
2215 2216 2217 2218 2219 2220
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2221
		size = posix_acl_xattr_size(acl->a_count);
2222 2223 2224
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2225 2226
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2227
			goto getout;
2228
		size = error;
2229 2230 2231
	} else
		size = 0;

2232 2233 2234
	error = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (error)
		goto getout;
2235
	if (size)
2236
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2237 2238 2239 2240
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2241
			error = vfs_removexattr(fhp->fh_dentry, name);
2242 2243 2244 2245
			if (error == -ENODATA)
				error = 0;
		}
	}
2246
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
2247 2248 2249 2250 2251 2252

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