vfs.c 49.5 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
struct raparm_hbucket {
	struct raparms		*pb_head;
	spinlock_t		pb_lock;
} ____cacheline_aligned_in_smp;

L
Linus Torvalds 已提交
86
static struct raparms *		raparml;
87 88 89 90
#define RAPARM_HASH_BITS	4
#define RAPARM_HASH_SIZE	(1<<RAPARM_HASH_BITS)
#define RAPARM_HASH_MASK	(RAPARM_HASH_SIZE-1)
static struct raparm_hbucket	raparm_hash[RAPARM_HASH_SIZE];
L
Linus Torvalds 已提交
91 92 93 94

/* 
 * Called from nfsd_lookup and encode_dirent. Check if we have crossed 
 * a mount point.
95
 * Returns -EAGAIN or -ETIMEDOUT leaving *dpp and *expp unchanged,
L
Linus Torvalds 已提交
96 97 98 99 100 101 102 103
 *  or nfs_ok having possibly changed *dpp and *expp
 */
int
nfsd_cross_mnt(struct svc_rqst *rqstp, struct dentry **dpp, 
		        struct svc_export **expp)
{
	struct svc_export *exp = *expp, *exp2 = NULL;
	struct dentry *dentry = *dpp;
104
	struct vfsmount *mnt = mntget(exp->ex_path.mnt);
L
Linus Torvalds 已提交
105
	struct dentry *mounts = dget(dentry);
106
	int err = 0;
L
Linus Torvalds 已提交
107 108 109

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

110
	exp2 = rqst_exp_get_by_name(rqstp, mnt, mounts);
L
Linus Torvalds 已提交
111
	if (IS_ERR(exp2)) {
112 113
		if (PTR_ERR(exp2) != -ENOENT)
			err = PTR_ERR(exp2);
L
Linus Torvalds 已提交
114 115 116 117
		dput(mounts);
		mntput(mnt);
		goto out;
	}
118
	if ((exp->ex_flags & NFSEXP_CROSSMOUNT) || EX_NOHIDE(exp2)) {
L
Linus Torvalds 已提交
119 120 121 122 123 124
		/* successfully crossed mount point */
		exp_put(exp);
		*expp = exp2;
		dput(dentry);
		*dpp = mounts;
	} else {
125
		exp_put(exp2);
L
Linus Torvalds 已提交
126 127 128 129 130 131 132
		dput(mounts);
	}
	mntput(mnt);
out:
	return err;
}

133
__be32
134
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
135
		   const char *name, unsigned int len,
136
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
137 138 139 140
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
141 142
	__be32			err;
	int			host_err;
L
Linus Torvalds 已提交
143 144 145 146

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

	/* Obtain dentry and export. */
M
Miklos Szeredi 已提交
147
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
148 149 150 151 152 153 154 155 156 157 158
	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);
159
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
160
			dentry = dget_parent(dparent);
161
		else if (!EX_NOHIDE(exp))
L
Linus Torvalds 已提交
162 163 164 165 166
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
			struct svc_export *exp2 = NULL;
			struct dentry *dp;
167
			struct vfsmount *mnt = mntget(exp->ex_path.mnt);
L
Linus Torvalds 已提交
168 169 170 171 172 173 174
			dentry = dget(dparent);
			while(dentry == mnt->mnt_root && follow_up(&mnt, &dentry))
				;
			dp = dget_parent(dentry);
			dput(dentry);
			dentry = dp;

175
			exp2 = rqst_exp_parent(rqstp, mnt, dentry);
176 177 178 179
			if (PTR_ERR(exp2) == -ENOENT) {
				dput(dentry);
				dentry = dget(dparent);
			} else if (IS_ERR(exp2)) {
180
				host_err = PTR_ERR(exp2);
L
Linus Torvalds 已提交
181 182 183 184 185 186 187 188 189 190 191 192
				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);
193
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
194 195 196 197 198 199
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
		if (d_mountpoint(dentry)) {
200
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
201 202 203 204 205
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
	*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,
229
				unsigned int len, struct svc_fh *resfh)
230 231 232 233 234 235 236 237
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
238 239 240
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
241 242 243 244 245 246 247
	/*
	 * 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;
248
out:
L
Linus Torvalds 已提交
249 250 251 252 253
	dput(dentry);
	exp_put(exp);
	return err;
}

254

L
Linus Torvalds 已提交
255 256 257 258
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
259
__be32
L
Linus Torvalds 已提交
260 261 262 263 264
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
265
	int		accmode = NFSD_MAY_SATTR;
L
Linus Torvalds 已提交
266
	int		ftype = 0;
267 268
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
269 270 271
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
272
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
273 274 275 276 277
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
278
	if (err)
L
Linus Torvalds 已提交
279 280 281 282 283
		goto out;

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

284 285 286 287 288 289 290
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

291 292
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
293 294 295 296 297 298 299 300 301 302 303 304
	 * 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)
305 306 307 308 309 310 311 312
	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
313
		 */
314 315 316 317 318 319 320 321 322 323 324 325
		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
326 327
	}
	    
328 329 330 331
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
332 333
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
334 335
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
336 337 338 339 340 341 342 343
			if (err)
				goto out;
		}

		/*
		 * If we are changing the size of the file, then
		 * we need to break all leases.
		 */
344 345 346 347
		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
348 349
			goto out_nfserr;

350 351
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
352 353 354
			goto out_nfserr;

		size_change = 1;
355 356
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
357 358 359 360 361 362
			put_write_access(inode);
			goto out_nfserr;
		}
		DQUOT_INIT(inode);
	}

363
	/* sanitize the mode change */
L
Linus Torvalds 已提交
364 365
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
366
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
367 368 369 370 371 372 373 374
	}

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

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
		fh_lock(fhp);
391 392
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
393 394 395 396 397 398 399 400 401 402 403
		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:
404
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
405 406 407
	goto out;
}

408 409 410 411 412 413 414 415 416 417
#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;

	buflen = vfs_getxattr(dentry, key, NULL, 0);
	if (buflen <= 0)
		return buflen;
L
Linus Torvalds 已提交
418

419 420 421 422
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

423
	return vfs_getxattr(dentry, key, *buf, buflen);
424 425 426 427
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
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;
	}

448
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
449 450 451 452 453
out:
	kfree(buf);
	return error;
}

454
__be32
L
Linus Torvalds 已提交
455 456 457
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
458 459
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
460 461 462 463 464 465
	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
466
	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
467
	if (error)
468
		return error;
L
Linus Torvalds 已提交
469 470 471 472 473 474

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

475 476
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
477
		return nfserr_attrnotsupp;
478
	} else if (host_error < 0)
L
Linus Torvalds 已提交
479 480
		goto out_nfserr;

481 482
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
483
		goto out_release;
L
Linus Torvalds 已提交
484

485
	if (S_ISDIR(inode->i_mode))
486
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
487

488
out_release:
L
Linus Torvalds 已提交
489 490 491
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
492
	if (host_error == -EOPNOTSUPP)
493
		return nfserr_attrnotsupp;
494
	else
495
		return nfserrno(host_error);
L
Linus Torvalds 已提交
496 497 498 499 500
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
501
	void *buf = NULL;
L
Linus Torvalds 已提交
502
	struct posix_acl *pacl = NULL;
503
	int buflen;
L
Linus Torvalds 已提交
504

505 506 507 508 509
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
510 511 512 513 514 515 516 517 518 519 520 521 522 523

	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;

524
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
525 526 527 528 529 530 531 532 533
	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)) {
534
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566
		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
567 568 569 570
    {	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
571 572 573 574 575

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
576 577 578 579 580
    {	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
581 582 583 584 585 586 587 588 589 590 591 592

    {	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
593 594 595 596
    {	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
597 598 599 600

    {	0,			0				}
};

601
__be32
L
Linus Torvalds 已提交
602 603 604 605 606 607
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;
608
	__be32			error;
L
Linus Torvalds 已提交
609

M
Miklos Szeredi 已提交
610
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
	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) {
628
			__be32 err2;
L
Linus Torvalds 已提交
629 630 631

			sresult |= map->access;

632
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
			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
 */
667
__be32
L
Linus Torvalds 已提交
668 669 670 671 672
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
			int access, struct file **filp)
{
	struct dentry	*dentry;
	struct inode	*inode;
673 674 675
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
676 677 678 679 680 681

	/*
	 * 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
682
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
683 684 685 686 687 688 689 690 691 692
	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
693
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
694
		goto out;
J
J. Bruce Fields 已提交
695 696 697 698 699 700
	/*
	 * 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
701 702 703 704 705 706 707 708 709
		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
710
	host_err = break_lease(inode, O_NONBLOCK | ((access & NFSD_MAY_WRITE) ? FMODE_WRITE : 0));
711 712 713
	if (host_err == -EWOULDBLOCK)
		host_err = -ETIMEDOUT;
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
714 715
		goto out_nfserr;

M
Miklos Szeredi 已提交
716 717
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
718 719 720
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
721 722 723

		DQUOT_INIT(inode);
	}
724 725
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
				flags);
L
Linus Torvalds 已提交
726
	if (IS_ERR(*filp))
727
		host_err = PTR_ERR(*filp);
L
Linus Torvalds 已提交
728
out_nfserr:
729
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
out:
	return err;
}

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

/*
 * Sync a file
 * As this calls fsync (not fdatasync) there is no need for a write_inode
 * after it.
 */
748
static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
749
			      const struct file_operations *fop)
L
Linus Torvalds 已提交
750 751 752
{
	struct inode *inode = dp->d_inode;
	int (*fsync) (struct file *, struct dentry *, int);
753
	int err;
L
Linus Torvalds 已提交
754

755 756 757 758 759
	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);
760

761
	return err;
L
Linus Torvalds 已提交
762 763 764
}
	

765
static int
L
Linus Torvalds 已提交
766 767
nfsd_sync(struct file *filp)
{
768
        int err;
769 770
	struct inode *inode = filp->f_path.dentry->d_inode;
	dprintk("nfsd: sync file %s\n", filp->f_path.dentry->d_name.name);
771
	mutex_lock(&inode->i_mutex);
772
	err=nfsd_dosync(filp, filp->f_path.dentry, filp->f_op);
773
	mutex_unlock(&inode->i_mutex);
774 775

	return err;
L
Linus Torvalds 已提交
776 777
}

778
int
L
Linus Torvalds 已提交
779 780
nfsd_sync_dir(struct dentry *dp)
{
781
	return nfsd_dosync(NULL, dp, dp->d_inode->i_fop);
L
Linus Torvalds 已提交
782 783 784 785 786 787 788 789 790 791 792 793
}

/*
 * 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;
794 795 796 797 798
	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
799

800 801
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
802 803 804 805 806 807 808 809
		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) {	
810
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
811 812 813 814 815 816 817
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
818
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
819
found:
820
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
821
		*rap = ra->p_next;
822 823
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
824 825 826
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
827
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
828 829 830 831
	return ra;
}

/*
832 833 834
 * 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
835 836
 */
static int
837 838
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
839
{
840
	struct svc_rqst *rqstp = sd->u.data;
841
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
842 843 844
	struct page *page = buf->page;
	size_t size;
	int ret;
L
Linus Torvalds 已提交
845

846
	ret = buf->ops->confirm(pipe, buf);
847 848 849 850
	if (unlikely(ret))
		return ret;

	size = sd->len;
L
Linus Torvalds 已提交
851 852 853

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
854 855 856
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
857
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
858
		rqstp->rq_res.page_len = size;
859
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
860
		get_page(page);
861 862
		if (*pp)
			put_page(*pp);
863 864
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
865
		rqstp->rq_res.page_len += size;
866
	} else
L
Linus Torvalds 已提交
867 868 869 870 871
		rqstp->rq_res.page_len += size;

	return size;
}

872 873 874 875 876 877
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

878 879 880 881 882 883 884 885 886
static inline int svc_msnfs(struct svc_fh *ffhp)
{
#ifdef MSNFS
	return (ffhp->fh_export->ex_flags & NFSEXP_MSNFS);
#else
	return 0;
#endif
}

887
static __be32
L
Linus Torvalds 已提交
888 889 890 891 892 893
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;
894 895
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
896 897

	err = nfserr_perm;
898
	inode = file->f_path.dentry->d_inode;
899 900

	if (svc_msnfs(fhp) && !lock_may_read(inode, offset, *count))
L
Linus Torvalds 已提交
901 902 903 904 905 906 907 908
		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;

909 910 911 912 913 914 915 916
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

917
		rqstp->rq_resused = 1;
918
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
919 920 921
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
922
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
923 924 925 926 927
		set_fs(oldfs);
	}

	/* Write back readahead params */
	if (ra) {
928 929
		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
		spin_lock(&rab->pb_lock);
L
Linus Torvalds 已提交
930 931 932
		ra->p_ra = file->f_ra;
		ra->p_set = 1;
		ra->p_count--;
933
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
934 935
	}

936 937 938
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
939
		err = 0;
940
		fsnotify_access(file->f_path.dentry);
L
Linus Torvalds 已提交
941
	} else 
942
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
943 944 945 946
out:
	return err;
}

947 948 949
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
950
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
951

952
	mutex_lock(&dentry->d_inode->i_mutex);
953
	notify_change(dentry, &ia);
954
	mutex_unlock(&dentry->d_inode->i_mutex);
955 956
}

957
static __be32
L
Linus Torvalds 已提交
958 959 960 961 962 963 964 965
nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
				loff_t offset, struct kvec *vec, int vlen,
	   			unsigned long cnt, int *stablep)
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	struct inode		*inode;
	mm_segment_t		oldfs;
966 967
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
968 969
	int			stable = *stablep;

970
#ifdef MSNFS
L
Linus Torvalds 已提交
971 972 973
	err = nfserr_perm;

	if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
974
		(!lock_may_write(file->f_path.dentry->d_inode, offset, cnt)))
L
Linus Torvalds 已提交
975 976 977
		goto out;
#endif

978
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
979 980 981 982 983 984 985 986 987 988 989 990
	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.
	 */

991
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
992 993 994 995 996 997 998 999 1000 1001 1002
	       stable = 2;
	       *stablep = 2; /* FILE_SYNC */
	}

	if (!EX_ISSYNC(exp))
		stable = 0;
	if (stable && !EX_WGATHER(exp))
		file->f_flags |= O_SYNC;

	/* Write the data. */
	oldfs = get_fs(); set_fs(KERNEL_DS);
1003
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1004
	set_fs(oldfs);
1005
	if (host_err >= 0) {
L
Linus Torvalds 已提交
1006
		nfsdstats.io_write += cnt;
1007
		fsnotify_modify(file->f_path.dentry);
L
Linus Torvalds 已提交
1008 1009 1010
	}

	/* clear setuid/setgid flag after write */
1011
	if (host_err >= 0 && (inode->i_mode & (S_ISUID | S_ISGID)))
1012
		kill_suid(dentry);
L
Linus Torvalds 已提交
1013

1014
	if (host_err >= 0 && stable) {
L
Linus Torvalds 已提交
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
		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)) {
1033
				dprintk("nfsd: write defer %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1034
				msleep(10);
1035
				dprintk("nfsd: write resume %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1036 1037 1038
			}

			if (inode->i_state & I_DIRTY) {
1039
				dprintk("nfsd: write sync %d\n", task_pid_nr(current));
1040
				host_err=nfsd_sync(file);
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049
			}
#if 0
			wake_up(&inode->i_wait);
#endif
		}
		last_ino = inode->i_ino;
		last_dev = inode->i_sb->s_dev;
	}

1050 1051
	dprintk("nfsd: write complete host_err=%d\n", host_err);
	if (host_err >= 0)
L
Linus Torvalds 已提交
1052 1053
		err = 0;
	else 
1054
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1055 1056 1057 1058 1059 1060 1061 1062 1063
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
 */
1064
__be32
L
Linus Torvalds 已提交
1065 1066 1067 1068
nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1069
	__be32		err;
L
Linus Torvalds 已提交
1070 1071

	if (file) {
1072
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1073
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1074 1075 1076 1077
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
	} else {
M
Miklos Szeredi 已提交
1078
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
		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
 */
1093
__be32
L
Linus Torvalds 已提交
1094 1095 1096 1097
nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
		loff_t offset, struct kvec *vec, int vlen, unsigned long cnt,
		int *stablep)
{
1098
	__be32			err = 0;
L
Linus Torvalds 已提交
1099 1100

	if (file) {
1101
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1102
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1103 1104 1105 1106 1107
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1108
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129
		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.
 */
1130
__be32
L
Linus Torvalds 已提交
1131 1132 1133 1134
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1135
	__be32		err;
L
Linus Torvalds 已提交
1136 1137 1138 1139

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

M
Miklos Szeredi 已提交
1140 1141
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
	if (err)
L
Linus Torvalds 已提交
1142 1143 1144
		return err;
	if (EX_ISSYNC(fhp->fh_export)) {
		if (file->f_op && file->f_op->fsync) {
1145
			err = nfserrno(nfsd_sync(file));
L
Linus Torvalds 已提交
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
		} else {
			err = nfserr_notsupp;
		}
	}

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

A
Adrian Bunk 已提交
1156
static __be32
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175
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:
	 */
	if (current->fsuid != 0)
		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
	if (iap->ia_valid)
		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
	return 0;
}

L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183
/*
 * 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
 */
1184
__be32
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190
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;
1191
	__be32		err;
1192
	__be32		err2;
1193
	int		host_err;
L
Linus Torvalds 已提交
1194 1195 1196 1197 1198 1199 1200 1201

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

M
Miklos Szeredi 已提交
1202
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
	if (err)
		goto out;

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

	err = nfserr_notdir;
	if(!dirp->i_op || !dirp->i_op->lookup)
		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
1218
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1219
		dchild = lookup_one_len(fname, dentry, flen);
1220
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234
		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
1235
			err = nfserr_io;
L
Linus Torvalds 已提交
1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
			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;

1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
	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
1264 1265 1266
	/*
	 * Get the dir op function pointer.
	 */
1267
	err = 0;
L
Linus Torvalds 已提交
1268 1269
	switch (type) {
	case S_IFREG:
1270
		host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
L
Linus Torvalds 已提交
1271 1272
		break;
	case S_IFDIR:
1273
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
1280 1281
		break;
	}
1282 1283
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1284
		goto out_nfserr;
1285
	}
L
Linus Torvalds 已提交
1286 1287

	if (EX_ISSYNC(fhp->fh_export)) {
1288
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1289 1290 1291
		write_inode_now(dchild->d_inode, 1);
	}

1292 1293 1294
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1295
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
1308 1309 1310 1311 1312 1313 1314
	goto out;
}

#ifdef CONFIG_NFSD_V3
/*
 * NFSv3 version of nfsd_create
 */
1315
__be32
L
Linus Torvalds 已提交
1316 1317 1318
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,
1319
	        int *truncp, int *created)
L
Linus Torvalds 已提交
1320 1321 1322
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1323
	__be32		err;
1324
	__be32		err2;
1325
	int		host_err;
L
Linus Torvalds 已提交
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335
	__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
1336
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347
	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;
	if(!dirp->i_op || !dirp->i_op->lookup)
		goto out;
P
Peter Zijlstra 已提交
1348
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1349 1350 1351 1352 1353

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1354
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1355 1356 1357 1358 1359 1360 1361 1362
	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) {
1363
		/* solaris7 gets confused (bugid 4218508) if these have
1364 1365 1366 1367
		 * 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
1368 1369 1370 1371 1372
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1373 1374 1375
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
	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;
		}
1407
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1408 1409 1410
		goto out;
	}

1411
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1412 1413
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1414
		goto out_nfserr;
1415
	}
1416 1417
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1418 1419

	if (EX_ISSYNC(fhp->fh_export)) {
1420
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1421 1422 1423 1424
		/* setattr will sync the child (or not) */
	}

	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1425
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1426
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1427
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1428 1429 1430 1431 1432 1433 1434 1435
		/* 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:
1436 1437 1438
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1439

1440
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
1441 1442 1443 1444 1445
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1446 1447 1448 1449 1450 1451 1452 1453

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1454
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1455 1456 1457 1458 1459 1460 1461 1462 1463
	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
 */
1464
__be32
L
Linus Torvalds 已提交
1465 1466 1467 1468 1469
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1470 1471
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1472

M
Miklos Szeredi 已提交
1473
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
	if (err)
		goto out;

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

	err = nfserr_inval;
	if (!inode->i_op || !inode->i_op->readlink)
		goto out;

1484
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1485 1486 1487 1488 1489
	/* 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);
1490
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1491 1492
	set_fs(oldfs);

1493
	if (host_err < 0)
L
Linus Torvalds 已提交
1494
		goto out_nfserr;
1495
	*lenp = host_err;
L
Linus Torvalds 已提交
1496 1497 1498 1499 1500
	err = 0;
out:
	return err;

out_nfserr:
1501
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1502 1503 1504 1505 1506 1507 1508
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1509
__be32
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514 1515 1516
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;
1517 1518
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1519 1520 1521 1522 1523 1524 1525 1526

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

M
Miklos Szeredi 已提交
1527
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1528 1529 1530 1531 1532
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1533
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1534 1535 1536
	if (IS_ERR(dnew))
		goto out_nfserr;

1537 1538 1539 1540
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1541 1542 1543
	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
1544
			host_err = -ENOMEM;
L
Linus Torvalds 已提交
1545 1546 1547
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
1548
			host_err = vfs_symlink(dentry->d_inode, dnew, path_alloced);
L
Linus Torvalds 已提交
1549 1550 1551
			kfree(path_alloced);
		}
	} else
1552
		host_err = vfs_symlink(dentry->d_inode, dnew, path);
L
Linus Torvalds 已提交
1553

1554
	if (!host_err) {
L
Linus Torvalds 已提交
1555
		if (EX_ISSYNC(fhp->fh_export))
1556 1557 1558
			host_err = nfsd_sync_dir(dentry);
	}
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1559 1560
	fh_unlock(fhp);

1561 1562
	mnt_drop_write(fhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1563 1564 1565 1566 1567 1568 1569
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1570
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1571 1572 1573 1574 1575 1576 1577
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1578
__be32
L
Linus Torvalds 已提交
1579 1580 1581 1582 1583
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;
1584 1585
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1586

M
Miklos Szeredi 已提交
1587
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1588 1589
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1590
	err = fh_verify(rqstp, tfhp, -S_IFDIR, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	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
1601
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1602 1603 1604 1605
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1606
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1607 1608 1609 1610 1611 1612
	if (IS_ERR(dnew))
		goto out_nfserr;

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

1613 1614 1615 1616 1617
	host_err = mnt_want_write(tfhp->fh_export->ex_path.mnt);
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1618 1619
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
L
Linus Torvalds 已提交
1620
		if (EX_ISSYNC(ffhp->fh_export)) {
1621
			err = nfserrno(nfsd_sync_dir(ddir));
L
Linus Torvalds 已提交
1622 1623
			write_inode_now(dest, 1);
		}
1624
		err = 0;
L
Linus Torvalds 已提交
1625
	} else {
1626
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1627 1628
			err = nfserr_acces;
		else
1629
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1630
	}
1631 1632
	mnt_drop_write(tfhp->fh_export->ex_path.mnt);
out_dput:
L
Linus Torvalds 已提交
1633
	dput(dnew);
1634 1635
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1636 1637 1638 1639
out:
	return err;

out_nfserr:
1640
	err = nfserrno(host_err);
1641
	goto out_unlock;
L
Linus Torvalds 已提交
1642 1643 1644 1645 1646 1647
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1648
__be32
L
Linus Torvalds 已提交
1649 1650 1651 1652 1653
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;
1654 1655
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1656

M
Miklos Szeredi 已提交
1657
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1658 1659
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1660
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
	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;
1671
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
		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);
1686
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1687 1688 1689
	if (IS_ERR(odentry))
		goto out_nfserr;

1690
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1691 1692
	if (!odentry->d_inode)
		goto out_dput_old;
1693
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1694 1695 1696 1697
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1698
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1699 1700
	if (IS_ERR(ndentry))
		goto out_dput_old;
1701
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1702 1703 1704
	if (ndentry == trap)
		goto out_dput_new;

1705
	if (svc_msnfs(ffhp) &&
L
Linus Torvalds 已提交
1706 1707
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
1708
			host_err = -EPERM;
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
			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;

1719 1720 1721 1722 1723
	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
1724 1725
	}

1726 1727
	mnt_drop_write(ffhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1728 1729 1730 1731 1732
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1733
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751

	/* 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
 */
1752
__be32
L
Linus Torvalds 已提交
1753 1754 1755 1756 1757
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1758 1759
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1760 1761 1762 1763

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1764
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1765 1766 1767
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1768
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1769 1770 1771 1772
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1773
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785
	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;

1786 1787 1788 1789
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1790 1791 1792 1793
	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
1794
			host_err = -EPERM;
L
Linus Torvalds 已提交
1795 1796
		} else
#endif
1797
		host_err = vfs_unlink(dirp, rdentry);
L
Linus Torvalds 已提交
1798
	} else { /* It's RMDIR */
1799
		host_err = vfs_rmdir(dirp, rdentry);
L
Linus Torvalds 已提交
1800 1801 1802 1803
	}

	dput(rdentry);

1804
	if (host_err)
1805
		goto out_drop;
1806 1807
	if (EX_ISSYNC(fhp->fh_export))
		host_err = nfsd_sync_dir(dentry);
L
Linus Torvalds 已提交
1808

1809 1810
out_drop:
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1811
out_nfserr:
1812
	err = nfserrno(host_err);
1813 1814
out:
	return err;
L
Linus Torvalds 已提交
1815 1816 1817 1818 1819 1820
}

/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1821
__be32
L
Linus Torvalds 已提交
1822
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1823
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1824
{
1825 1826
	__be32		err;
	int 		host_err;
L
Linus Torvalds 已提交
1827 1828 1829
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
1830
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
	if (err)
		goto out;

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

	/*
	 * Read the directory entries. This silly loop is necessary because
	 * readdir() is not guaranteed to fill up the entire buffer, but
	 * may choose to do less.
	 */

	do {
		cdp->err = nfserr_eof; /* will be cleared on successful read */
1848
		host_err = vfs_readdir(file, func, cdp);
1849 1850 1851
	} while (host_err >=0 && cdp->err == nfs_ok);
	if (host_err)
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
	else
		err = cdp->err;
	*offsetp = vfs_llseek(file, 0, 1);

	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
 */
1868
__be32
L
Linus Torvalds 已提交
1869 1870
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat)
{
M
Miklos Szeredi 已提交
1871
	__be32 err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
1872
	if (!err && vfs_statfs(fhp->fh_dentry,stat))
L
Linus Torvalds 已提交
1873 1874 1875 1876
		err = nfserr_io;
	return err;
}

J
J. Bruce Fields 已提交
1877
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
1878
{
J
J. Bruce Fields 已提交
1879
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
1880 1881
}

L
Linus Torvalds 已提交
1882 1883 1884
/*
 * Check for a user's access permissions to this inode.
 */
1885
__be32
1886 1887
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
1888 1889 1890 1891
{
	struct inode	*inode = dentry->d_inode;
	int		err;

M
Miklos Szeredi 已提交
1892
	if (acc == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
1893 1894 1895 1896
		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
1897 1898 1899 1900 1901 1902 1903
		(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
1904 1905 1906
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
1907
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
1908 1909 1910 1911 1912 1913 1914 1915
	dprintk("      owner %d/%d user %d/%d\n",
		inode->i_uid, inode->i_gid, current->fsuid, current->fsgid);
#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
1916 1917
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
1918 1919
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
1920
				return nfserr_rofs;
M
Miklos Szeredi 已提交
1921
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
1922 1923
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
1924
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
1925 1926
		return nfserr_perm;

M
Miklos Szeredi 已提交
1927
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
1928 1929 1930 1931 1932 1933 1934
		/* 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
1935
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
	}
	/*
	 * 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
1951
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
L
Linus Torvalds 已提交
1952 1953 1954
	    inode->i_uid == current->fsuid)
		return 0;

M
Miklos Szeredi 已提交
1955
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
1956
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
1957 1958 1959

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
M
Miklos Szeredi 已提交
1960
	    acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE))
1961
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
1962 1963 1964 1965 1966 1967 1968

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
1969
	if (!raparml)
L
Linus Torvalds 已提交
1970 1971 1972
		return;
	dprintk("nfsd: freeing readahead buffers.\n");
	kfree(raparml);
1973
	raparml = NULL;
L
Linus Torvalds 已提交
1974 1975 1976 1977 1978 1979 1980 1981
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
1982 1983
	int	j = 0;
	int	nperbucket;
L
Linus Torvalds 已提交
1984

1985 1986

	if (raparml)
L
Linus Torvalds 已提交
1987
		return 0;
1988 1989
	if (cache_size < 2*RAPARM_HASH_SIZE)
		cache_size = 2*RAPARM_HASH_SIZE;
1990 1991 1992
	raparml = kcalloc(cache_size, sizeof(struct raparms), GFP_KERNEL);

	if (!raparml) {
L
Linus Torvalds 已提交
1993
		printk(KERN_WARNING
1994
			"nfsd: Could not allocate memory read-ahead cache.\n");
L
Linus Torvalds 已提交
1995 1996
		return -ENOMEM;
	}
1997 1998 1999 2000 2001 2002

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
	for (i = 0 ; i < RAPARM_HASH_SIZE ; i++) {
		raparm_hash[i].pb_head = NULL;
		spin_lock_init(&raparm_hash[i].pb_lock);
	}
2003
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
2004 2005 2006 2007 2008 2009 2010
	for (i = 0; i < cache_size - 1; i++) {
		if (i % nperbucket == 0)
			raparm_hash[j++].pb_head = raparml + i;
		if (i % nperbucket < nperbucket-1)
			raparml[i].p_next = raparml + i + 1;
	}

L
Linus Torvalds 已提交
2011 2012 2013
	nfsdstats.ra_size = cache_size;
	return 0;
}
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024

#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;

2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
	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:
2036 2037 2038
		return ERR_PTR(-EOPNOTSUPP);
	}

2039 2040 2041
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061

	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;

	if (!IS_POSIXACL(inode) || !inode->i_op ||
	    !inode->i_op->setxattr || !inode->i_op->removexattr)
		return -EOPNOTSUPP;
	switch(type) {
		case ACL_TYPE_ACCESS:
2062
			name = POSIX_ACL_XATTR_ACCESS;
2063 2064
			break;
		case ACL_TYPE_DEFAULT:
2065
			name = POSIX_ACL_XATTR_DEFAULT;
2066 2067 2068 2069 2070 2071
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2072
		size = posix_acl_xattr_size(acl->a_count);
2073 2074 2075
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2076 2077
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2078
			goto getout;
2079
		size = error;
2080 2081 2082
	} else
		size = 0;

2083 2084 2085
	error = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (error)
		goto getout;
2086
	if (size)
2087
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2088 2089 2090 2091
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2092
			error = vfs_removexattr(fhp->fh_dentry, name);
2093 2094 2095 2096
			if (error == -ENODATA)
				error = 0;
		}
	}
2097
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
2098 2099 2100 2101 2102 2103

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