vfs.c 49.7 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
118 119 120 121 122 123
		/* successfully crossed mount point */
		exp_put(exp);
		*expp = exp2;
		dput(dentry);
		*dpp = mounts;
	} else {
124
		exp_put(exp2);
L
Linus Torvalds 已提交
125 126 127 128 129 130 131
		dput(mounts);
	}
	mntput(mnt);
out:
	return err;
}

132
__be32
133
nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
134
		   const char *name, unsigned int len,
135
		   struct svc_export **exp_ret, struct dentry **dentry_ret)
L
Linus Torvalds 已提交
136 137 138 139
{
	struct svc_export	*exp;
	struct dentry		*dparent;
	struct dentry		*dentry;
140 141
	__be32			err;
	int			host_err;
L
Linus Torvalds 已提交
142 143 144 145

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

	/* Obtain dentry and export. */
M
Miklos Szeredi 已提交
146
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_EXEC);
L
Linus Torvalds 已提交
147 148 149 150 151 152 153 154 155 156 157
	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);
158
		else if (dparent != exp->ex_path.dentry)
L
Linus Torvalds 已提交
159
			dentry = dget_parent(dparent);
160
		else if (!EX_NOHIDE(exp))
L
Linus Torvalds 已提交
161 162 163 164 165
			dentry = dget(dparent); /* .. == . just like at / */
		else {
			/* checking mountpoint crossing is very different when stepping up */
			struct svc_export *exp2 = NULL;
			struct dentry *dp;
166
			struct vfsmount *mnt = mntget(exp->ex_path.mnt);
L
Linus Torvalds 已提交
167 168 169 170 171 172 173
			dentry = dget(dparent);
			while(dentry == mnt->mnt_root && follow_up(&mnt, &dentry))
				;
			dp = dget_parent(dentry);
			dput(dentry);
			dentry = dp;

174
			exp2 = rqst_exp_parent(rqstp, mnt, dentry);
175 176 177 178
			if (PTR_ERR(exp2) == -ENOENT) {
				dput(dentry);
				dentry = dget(dparent);
			} else if (IS_ERR(exp2)) {
179
				host_err = PTR_ERR(exp2);
L
Linus Torvalds 已提交
180 181 182 183 184 185 186 187 188 189 190 191
				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);
192
		host_err = PTR_ERR(dentry);
L
Linus Torvalds 已提交
193 194 195 196 197 198
		if (IS_ERR(dentry))
			goto out_nfserr;
		/*
		 * check if we have crossed a mount point ...
		 */
		if (d_mountpoint(dentry)) {
199
			if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) {
L
Linus Torvalds 已提交
200 201 202 203 204
				dput(dentry);
				goto out_nfserr;
			}
		}
	}
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
	*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,
228
				unsigned int len, struct svc_fh *resfh)
229 230 231 232 233 234 235 236
{
	struct svc_export	*exp;
	struct dentry		*dentry;
	__be32 err;

	err = nfsd_lookup_dentry(rqstp, fhp, name, len, &exp, &dentry);
	if (err)
		return err;
237 238 239
	err = check_nfsd_access(exp, rqstp);
	if (err)
		goto out;
L
Linus Torvalds 已提交
240 241 242 243 244 245 246
	/*
	 * 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;
247
out:
L
Linus Torvalds 已提交
248 249 250 251 252
	dput(dentry);
	exp_put(exp);
	return err;
}

253

L
Linus Torvalds 已提交
254 255 256 257
/*
 * Set various file attributes.
 * N.B. After this call fhp needs an fh_put
 */
258
__be32
L
Linus Torvalds 已提交
259 260 261 262 263
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
264
	int		accmode = NFSD_MAY_SATTR;
L
Linus Torvalds 已提交
265
	int		ftype = 0;
266 267
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
268 269 270
	int		size_change = 0;

	if (iap->ia_valid & (ATTR_ATIME | ATTR_MTIME | ATTR_SIZE))
M
Miklos Szeredi 已提交
271
		accmode |= NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
272 273 274 275 276
	if (iap->ia_valid & ATTR_SIZE)
		ftype = S_IFREG;

	/* Get inode */
	err = fh_verify(rqstp, fhp, ftype, accmode);
277
	if (err)
L
Linus Torvalds 已提交
278 279 280 281 282
		goto out;

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

283 284 285 286 287 288 289
	/* Ignore any mode updates on symlinks */
	if (S_ISLNK(inode->i_mode))
		iap->ia_valid &= ~ATTR_MODE;

	if (!iap->ia_valid)
		goto out;

290 291
	/*
	 * NFSv2 does not differentiate between "set-[ac]time-to-now"
L
Linus Torvalds 已提交
292 293 294 295 296 297 298 299 300 301 302 303
	 * 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)
304 305 306 307 308 309 310 311
	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
312
		 */
313 314 315 316 317 318 319 320 321 322 323 324
		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
325 326
	}
	    
327 328 329 330
	/*
	 * The size case is special.
	 * It changes the file as well as the attributes.
	 */
L
Linus Torvalds 已提交
331 332
	if (iap->ia_valid & ATTR_SIZE) {
		if (iap->ia_size < inode->i_size) {
M
Miklos Szeredi 已提交
333 334
			err = nfsd_permission(rqstp, fhp->fh_export, dentry,
					NFSD_MAY_TRUNC|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
335 336 337 338 339 340 341 342
			if (err)
				goto out;
		}

		/*
		 * If we are changing the size of the file, then
		 * we need to break all leases.
		 */
343 344 345 346
		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
347 348
			goto out_nfserr;

349 350
		host_err = get_write_access(inode);
		if (host_err)
L
Linus Torvalds 已提交
351 352 353
			goto out_nfserr;

		size_change = 1;
354 355
		host_err = locks_verify_truncate(inode, NULL, iap->ia_size);
		if (host_err) {
L
Linus Torvalds 已提交
356 357 358 359 360 361
			put_write_access(inode);
			goto out_nfserr;
		}
		DQUOT_INIT(inode);
	}

362
	/* sanitize the mode change */
L
Linus Torvalds 已提交
363 364
	if (iap->ia_valid & ATTR_MODE) {
		iap->ia_mode &= S_IALLUGO;
365
		iap->ia_mode |= (inode->i_mode & ~S_IALLUGO);
366 367 368 369 370 371 372 373
	}

	/* 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 */
374
			iap->ia_mode &= ~S_ISUID;
375 376 377 378 379
			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);
380
		}
L
Linus Torvalds 已提交
381 382 383 384 385 386 387 388 389
	}

	/* Change the attributes. */

	iap->ia_valid |= ATTR_CTIME;

	err = nfserr_notsync;
	if (!check_guard || guardtime == inode->i_ctime.tv_sec) {
		fh_lock(fhp);
390 391
		host_err = notify_change(dentry, iap);
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
392 393 394 395 396 397 398 399 400 401 402
		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:
403
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
404 405 406
	goto out;
}

407 408 409 410 411 412 413 414 415 416
#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
417

418 419 420 421
	*buf = kmalloc(buflen, GFP_KERNEL);
	if (!*buf)
		return -ENOMEM;

422
	return vfs_getxattr(dentry, key, *buf, buflen);
423 424 425 426
}
#endif

#if defined(CONFIG_NFSD_V4)
L
Linus Torvalds 已提交
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
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;
	}

447
	error = vfs_setxattr(dentry, key, buf, len, 0);
L
Linus Torvalds 已提交
448 449 450 451 452
out:
	kfree(buf);
	return error;
}

453
__be32
L
Linus Torvalds 已提交
454 455 456
nfsd4_set_nfs4_acl(struct svc_rqst *rqstp, struct svc_fh *fhp,
    struct nfs4_acl *acl)
{
457 458
	__be32 error;
	int host_error;
L
Linus Torvalds 已提交
459 460 461 462 463 464
	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
465
	error = fh_verify(rqstp, fhp, 0 /* S_IFREG */, NFSD_MAY_SATTR);
L
Linus Torvalds 已提交
466
	if (error)
467
		return error;
L
Linus Torvalds 已提交
468 469 470 471 472 473

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

474 475
	host_error = nfs4_acl_nfsv4_to_posix(acl, &pacl, &dpacl, flags);
	if (host_error == -EINVAL) {
476
		return nfserr_attrnotsupp;
477
	} else if (host_error < 0)
L
Linus Torvalds 已提交
478 479
		goto out_nfserr;

480 481
	host_error = set_nfsv4_acl_one(dentry, pacl, POSIX_ACL_XATTR_ACCESS);
	if (host_error < 0)
482
		goto out_release;
L
Linus Torvalds 已提交
483

484
	if (S_ISDIR(inode->i_mode))
485
		host_error = set_nfsv4_acl_one(dentry, dpacl, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
486

487
out_release:
L
Linus Torvalds 已提交
488 489 490
	posix_acl_release(pacl);
	posix_acl_release(dpacl);
out_nfserr:
491
	if (host_error == -EOPNOTSUPP)
492
		return nfserr_attrnotsupp;
493
	else
494
		return nfserrno(host_error);
L
Linus Torvalds 已提交
495 496 497 498 499
}

static struct posix_acl *
_get_posix_acl(struct dentry *dentry, char *key)
{
500
	void *buf = NULL;
L
Linus Torvalds 已提交
501
	struct posix_acl *pacl = NULL;
502
	int buflen;
L
Linus Torvalds 已提交
503

504 505 506 507 508
	buflen = nfsd_getxattr(dentry, key, &buf);
	if (!buflen)
		buflen = -ENODATA;
	if (buflen <= 0)
		return ERR_PTR(buflen);
L
Linus Torvalds 已提交
509 510 511 512 513 514 515 516 517 518 519 520 521 522

	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;

523
	pacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_ACCESS);
L
Linus Torvalds 已提交
524 525 526 527 528 529 530 531 532
	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)) {
533
		dpacl = _get_posix_acl(dentry, POSIX_ACL_XATTR_DEFAULT);
L
Linus Torvalds 已提交
534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565
		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
566 567 568 569
    {	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
570 571 572 573 574

    {	0,			0				}
};

static struct accessmap	nfs3_diraccess[] = {
M
Miklos Szeredi 已提交
575 576 577 578 579
    {	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
580 581 582 583 584 585 586 587 588 589 590 591

    {	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
592 593 594 595
    {	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
596 597 598 599

    {	0,			0				}
};

600
__be32
L
Linus Torvalds 已提交
601 602 603 604 605 606
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;
607
	__be32			error;
L
Linus Torvalds 已提交
608

M
Miklos Szeredi 已提交
609
	error = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
	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) {
627
			__be32 err2;
L
Linus Torvalds 已提交
628 629 630

			sresult |= map->access;

631
			err2 = nfsd_permission(rqstp, export, dentry, map->how);
L
Linus Torvalds 已提交
632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665
			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
 */
666
__be32
L
Linus Torvalds 已提交
667 668 669 670 671
nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
			int access, struct file **filp)
{
	struct dentry	*dentry;
	struct inode	*inode;
672 673 674
	int		flags = O_RDONLY|O_LARGEFILE;
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
675 676 677 678 679 680

	/*
	 * 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
681
	err = fh_verify(rqstp, fhp, type, access | NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
682 683 684 685 686 687 688 689 690 691
	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
692
	if (IS_APPEND(inode) && (access & NFSD_MAY_WRITE))
L
Linus Torvalds 已提交
693
		goto out;
J
J. Bruce Fields 已提交
694 695 696 697 698 699
	/*
	 * 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
700 701 702 703 704 705 706 707 708
		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
709
	host_err = break_lease(inode, O_NONBLOCK | ((access & NFSD_MAY_WRITE) ? FMODE_WRITE : 0));
710 711 712
	if (host_err == -EWOULDBLOCK)
		host_err = -ETIMEDOUT;
	if (host_err) /* NOMEM or WOULDBLOCK */
L
Linus Torvalds 已提交
713 714
		goto out_nfserr;

M
Miklos Szeredi 已提交
715 716
	if (access & NFSD_MAY_WRITE) {
		if (access & NFSD_MAY_READ)
717 718 719
			flags = O_RDWR|O_LARGEFILE;
		else
			flags = O_WRONLY|O_LARGEFILE;
L
Linus Torvalds 已提交
720 721 722

		DQUOT_INIT(inode);
	}
723 724
	*filp = dentry_open(dget(dentry), mntget(fhp->fh_export->ex_path.mnt),
				flags);
L
Linus Torvalds 已提交
725
	if (IS_ERR(*filp))
726
		host_err = PTR_ERR(*filp);
L
Linus Torvalds 已提交
727
out_nfserr:
728
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
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.
 */
747
static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
748
			      const struct file_operations *fop)
L
Linus Torvalds 已提交
749 750 751
{
	struct inode *inode = dp->d_inode;
	int (*fsync) (struct file *, struct dentry *, int);
752
	int err;
L
Linus Torvalds 已提交
753

754 755 756 757 758
	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);
759

760
	return err;
L
Linus Torvalds 已提交
761 762 763
}
	

764
static int
L
Linus Torvalds 已提交
765 766
nfsd_sync(struct file *filp)
{
767
        int err;
768 769
	struct inode *inode = filp->f_path.dentry->d_inode;
	dprintk("nfsd: sync file %s\n", filp->f_path.dentry->d_name.name);
770
	mutex_lock(&inode->i_mutex);
771
	err=nfsd_dosync(filp, filp->f_path.dentry, filp->f_op);
772
	mutex_unlock(&inode->i_mutex);
773 774

	return err;
L
Linus Torvalds 已提交
775 776
}

777
int
L
Linus Torvalds 已提交
778 779
nfsd_sync_dir(struct dentry *dp)
{
780
	return nfsd_dosync(NULL, dp, dp->d_inode->i_fop);
L
Linus Torvalds 已提交
781 782 783 784 785 786 787 788 789 790 791 792
}

/*
 * 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;
793 794 795 796 797
	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
798

799 800
	spin_lock(&rab->pb_lock);
	for (rap = &rab->pb_head; (ra = *rap); rap = &ra->p_next) {
L
Linus Torvalds 已提交
801 802 803 804 805 806 807 808
		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) {	
809
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
810 811 812 813 814 815 816
		return NULL;
	}
	rap = frap;
	ra = *frap;
	ra->p_dev = dev;
	ra->p_ino = ino;
	ra->p_set = 0;
817
	ra->p_hindex = hash;
L
Linus Torvalds 已提交
818
found:
819
	if (rap != &rab->pb_head) {
L
Linus Torvalds 已提交
820
		*rap = ra->p_next;
821 822
		ra->p_next   = rab->pb_head;
		rab->pb_head = ra;
L
Linus Torvalds 已提交
823 824 825
	}
	ra->p_count++;
	nfsdstats.ra_depth[depth*10/nfsdstats.ra_size]++;
826
	spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
827 828 829 830
	return ra;
}

/*
831 832 833
 * 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
834 835
 */
static int
836 837
nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
		  struct splice_desc *sd)
L
Linus Torvalds 已提交
838
{
839
	struct svc_rqst *rqstp = sd->u.data;
840
	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
841 842 843
	struct page *page = buf->page;
	size_t size;
	int ret;
L
Linus Torvalds 已提交
844

845
	ret = buf->ops->confirm(pipe, buf);
846 847 848 849
	if (unlikely(ret))
		return ret;

	size = sd->len;
L
Linus Torvalds 已提交
850 851 852

	if (rqstp->rq_res.page_len == 0) {
		get_page(page);
853 854 855
		put_page(*pp);
		*pp = page;
		rqstp->rq_resused++;
856
		rqstp->rq_res.page_base = buf->offset;
L
Linus Torvalds 已提交
857
		rqstp->rq_res.page_len = size;
858
	} else if (page != pp[-1]) {
L
Linus Torvalds 已提交
859
		get_page(page);
860 861
		if (*pp)
			put_page(*pp);
862 863
		*pp = page;
		rqstp->rq_resused++;
L
Linus Torvalds 已提交
864
		rqstp->rq_res.page_len += size;
865
	} else
L
Linus Torvalds 已提交
866 867 868 869 870
		rqstp->rq_res.page_len += size;

	return size;
}

871 872 873 874 875 876
static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
				    struct splice_desc *sd)
{
	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
}

877 878 879 880 881 882 883 884 885
static inline int svc_msnfs(struct svc_fh *ffhp)
{
#ifdef MSNFS
	return (ffhp->fh_export->ex_flags & NFSEXP_MSNFS);
#else
	return 0;
#endif
}

886
static __be32
L
Linus Torvalds 已提交
887 888 889 890 891 892
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;
893 894
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
895 896

	err = nfserr_perm;
897
	inode = file->f_path.dentry->d_inode;
898 899

	if (svc_msnfs(fhp) && !lock_may_read(inode, offset, *count))
L
Linus Torvalds 已提交
900 901 902 903 904 905 906 907
		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;

908 909 910 911 912 913 914 915
	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
		struct splice_desc sd = {
			.len		= 0,
			.total_len	= *count,
			.pos		= offset,
			.u.data		= rqstp,
		};

916
		rqstp->rq_resused = 1;
917
		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
L
Linus Torvalds 已提交
918 919 920
	} else {
		oldfs = get_fs();
		set_fs(KERNEL_DS);
921
		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
922 923 924 925 926
		set_fs(oldfs);
	}

	/* Write back readahead params */
	if (ra) {
927 928
		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
		spin_lock(&rab->pb_lock);
L
Linus Torvalds 已提交
929 930 931
		ra->p_ra = file->f_ra;
		ra->p_set = 1;
		ra->p_count--;
932
		spin_unlock(&rab->pb_lock);
L
Linus Torvalds 已提交
933 934
	}

935 936 937
	if (host_err >= 0) {
		nfsdstats.io_read += host_err;
		*count = host_err;
L
Linus Torvalds 已提交
938
		err = 0;
939
		fsnotify_access(file->f_path.dentry);
L
Linus Torvalds 已提交
940
	} else 
941
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
942 943 944 945
out:
	return err;
}

946 947 948
static void kill_suid(struct dentry *dentry)
{
	struct iattr	ia;
949
	ia.ia_valid = ATTR_KILL_SUID | ATTR_KILL_SGID | ATTR_KILL_PRIV;
950

951
	mutex_lock(&dentry->d_inode->i_mutex);
952
	notify_change(dentry, &ia);
953
	mutex_unlock(&dentry->d_inode->i_mutex);
954 955
}

956
static __be32
L
Linus Torvalds 已提交
957 958 959 960 961 962 963 964
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;
965 966
	__be32			err = 0;
	int			host_err;
L
Linus Torvalds 已提交
967 968
	int			stable = *stablep;

969
#ifdef MSNFS
L
Linus Torvalds 已提交
970 971 972
	err = nfserr_perm;

	if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
973
		(!lock_may_write(file->f_path.dentry->d_inode, offset, cnt)))
L
Linus Torvalds 已提交
974 975 976
		goto out;
#endif

977
	dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
978 979 980 981 982 983 984 985 986 987 988 989
	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.
	 */

990
	if (!file->f_op->fsync) {/* COMMIT3 cannot work */
L
Linus Torvalds 已提交
991 992 993 994 995 996 997 998 999 1000 1001
	       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);
1002
	host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset);
L
Linus Torvalds 已提交
1003
	set_fs(oldfs);
1004
	if (host_err >= 0) {
L
Linus Torvalds 已提交
1005
		nfsdstats.io_write += cnt;
1006
		fsnotify_modify(file->f_path.dentry);
L
Linus Torvalds 已提交
1007 1008 1009
	}

	/* clear setuid/setgid flag after write */
1010
	if (host_err >= 0 && (inode->i_mode & (S_ISUID | S_ISGID)))
1011
		kill_suid(dentry);
L
Linus Torvalds 已提交
1012

1013
	if (host_err >= 0 && stable) {
L
Linus Torvalds 已提交
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
		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)) {
1032
				dprintk("nfsd: write defer %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1033
				msleep(10);
1034
				dprintk("nfsd: write resume %d\n", task_pid_nr(current));
L
Linus Torvalds 已提交
1035 1036 1037
			}

			if (inode->i_state & I_DIRTY) {
1038
				dprintk("nfsd: write sync %d\n", task_pid_nr(current));
1039
				host_err=nfsd_sync(file);
L
Linus Torvalds 已提交
1040 1041 1042 1043 1044 1045 1046 1047 1048
			}
#if 0
			wake_up(&inode->i_wait);
#endif
		}
		last_ino = inode->i_ino;
		last_dev = inode->i_sb->s_dev;
	}

1049 1050
	dprintk("nfsd: write complete host_err=%d\n", host_err);
	if (host_err >= 0)
L
Linus Torvalds 已提交
1051 1052
		err = 0;
	else 
1053
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062
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
 */
1063
__be32
L
Linus Torvalds 已提交
1064 1065 1066 1067
nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
		loff_t offset, struct kvec *vec, int vlen,
		unsigned long *count)
{
1068
	__be32		err;
L
Linus Torvalds 已提交
1069 1070

	if (file) {
1071
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1072
				NFSD_MAY_READ|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1073 1074 1075 1076
		if (err)
			goto out;
		err = nfsd_vfs_read(rqstp, fhp, file, offset, vec, vlen, count);
	} else {
M
Miklos Szeredi 已提交
1077
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
		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
 */
1092
__be32
L
Linus Torvalds 已提交
1093 1094 1095 1096
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)
{
1097
	__be32			err = 0;
L
Linus Torvalds 已提交
1098 1099

	if (file) {
1100
		err = nfsd_permission(rqstp, fhp->fh_export, fhp->fh_dentry,
M
Miklos Szeredi 已提交
1101
				NFSD_MAY_WRITE|NFSD_MAY_OWNER_OVERRIDE);
L
Linus Torvalds 已提交
1102 1103 1104 1105 1106
		if (err)
			goto out;
		err = nfsd_vfs_write(rqstp, fhp, file, offset, vec, vlen, cnt,
				stablep);
	} else {
M
Miklos Szeredi 已提交
1107
		err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
L
Linus Torvalds 已提交
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128
		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.
 */
1129
__be32
L
Linus Torvalds 已提交
1130 1131 1132 1133
nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
               loff_t offset, unsigned long count)
{
	struct file	*file;
1134
	__be32		err;
L
Linus Torvalds 已提交
1135 1136 1137 1138

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

M
Miklos Szeredi 已提交
1139 1140
	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_WRITE, &file);
	if (err)
L
Linus Torvalds 已提交
1141 1142 1143
		return err;
	if (EX_ISSYNC(fhp->fh_export)) {
		if (file->f_op && file->f_op->fsync) {
1144
			err = nfserrno(nfsd_sync(file));
L
Linus Torvalds 已提交
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154
		} else {
			err = nfserr_notsupp;
		}
	}

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

A
Adrian Bunk 已提交
1155
static __be32
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
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
1175 1176 1177 1178 1179 1180 1181 1182
/*
 * 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
 */
1183
__be32
L
Linus Torvalds 已提交
1184 1185 1186 1187 1188 1189
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;
1190
	__be32		err;
1191
	__be32		err2;
1192
	int		host_err;
L
Linus Torvalds 已提交
1193 1194 1195 1196 1197 1198 1199 1200

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

M
Miklos Szeredi 已提交
1201
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
	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
1217
		fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1218
		dchild = lookup_one_len(fname, dentry, flen);
1219
		host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
		if (IS_ERR(dchild))
			goto out_nfserr;
		err = fh_compose(resfhp, fhp->fh_export, dchild, fhp);
		if (err)
			goto out;
	} else {
		/* called from nfsd_proc_create */
		dchild = dget(resfhp->fh_dentry);
		if (!fhp->fh_locked) {
			/* not actually possible */
			printk(KERN_ERR
				"nfsd_create: parent %s/%s not locked!\n",
				dentry->d_parent->d_name.name,
				dentry->d_name.name);
A
Al Viro 已提交
1234
			err = nfserr_io;
L
Linus Torvalds 已提交
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
			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;

1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
	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
1263 1264 1265
	/*
	 * Get the dir op function pointer.
	 */
1266
	err = 0;
L
Linus Torvalds 已提交
1267 1268
	switch (type) {
	case S_IFREG:
1269
		host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
L
Linus Torvalds 已提交
1270 1271
		break;
	case S_IFDIR:
1272
		host_err = vfs_mkdir(dirp, dchild, iap->ia_mode);
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277
		break;
	case S_IFCHR:
	case S_IFBLK:
	case S_IFIFO:
	case S_IFSOCK:
1278
		host_err = vfs_mknod(dirp, dchild, iap->ia_mode, rdev);
L
Linus Torvalds 已提交
1279 1280
		break;
	}
1281 1282
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1283
		goto out_nfserr;
1284
	}
L
Linus Torvalds 已提交
1285 1286

	if (EX_ISSYNC(fhp->fh_export)) {
1287
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1288 1289 1290
		write_inode_now(dchild->d_inode, 1);
	}

1291 1292 1293
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1294
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
	/*
	 * 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:
1306
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1307 1308 1309 1310 1311 1312 1313
	goto out;
}

#ifdef CONFIG_NFSD_V3
/*
 * NFSv3 version of nfsd_create
 */
1314
__be32
L
Linus Torvalds 已提交
1315 1316 1317
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,
1318
	        int *truncp, int *created)
L
Linus Torvalds 已提交
1319 1320 1321
{
	struct dentry	*dentry, *dchild = NULL;
	struct inode	*dirp;
1322
	__be32		err;
1323
	__be32		err2;
1324
	int		host_err;
L
Linus Torvalds 已提交
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
	__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
1335
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
	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
1347
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1348 1349 1350 1351 1352

	/*
	 * Compose the response file handle.
	 */
	dchild = lookup_one_len(fname, dentry, flen);
1353
	host_err = PTR_ERR(dchild);
L
Linus Torvalds 已提交
1354 1355 1356 1357 1358 1359 1360 1361
	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) {
1362
		/* solaris7 gets confused (bugid 4218508) if these have
1363 1364 1365 1366
		 * 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
1367 1368 1369 1370 1371
		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
	}
	
1372 1373 1374
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;
L
Linus Torvalds 已提交
1375 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
	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;
		}
1406
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1407 1408 1409
		goto out;
	}

1410
	host_err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
1411 1412
	if (host_err < 0) {
		mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1413
		goto out_nfserr;
1414
	}
1415 1416
	if (created)
		*created = 1;
L
Linus Torvalds 已提交
1417 1418

	if (EX_ISSYNC(fhp->fh_export)) {
1419
		err = nfserrno(nfsd_sync_dir(dentry));
L
Linus Torvalds 已提交
1420 1421 1422 1423
		/* setattr will sync the child (or not) */
	}

	if (createmode == NFS3_CREATE_EXCLUSIVE) {
1424
		/* Cram the verifier into atime/mtime */
L
Linus Torvalds 已提交
1425
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
1426
			| ATTR_MTIME_SET|ATTR_ATIME_SET;
L
Linus Torvalds 已提交
1427 1428 1429 1430 1431 1432 1433 1434
		/* 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:
1435 1436 1437
	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
	if (err2)
		err = err2;
1438

1439
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
1440 1441 1442 1443 1444
	/*
	 * Update the filehandle to get the new inode info.
	 */
	if (!err)
		err = fh_update(resfhp);
L
Linus Torvalds 已提交
1445 1446 1447 1448 1449 1450 1451 1452

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
1453
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1454 1455 1456 1457 1458 1459 1460 1461 1462
	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
 */
1463
__be32
L
Linus Torvalds 已提交
1464 1465 1466 1467 1468
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
1469 1470
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1471

M
Miklos Szeredi 已提交
1472
	err = fh_verify(rqstp, fhp, S_IFLNK, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1473 1474 1475 1476 1477 1478 1479 1480 1481 1482
	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;

1483
	touch_atime(fhp->fh_export->ex_path.mnt, dentry);
L
Linus Torvalds 已提交
1484 1485 1486 1487 1488
	/* 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);
1489
	host_err = inode->i_op->readlink(dentry, buf, *lenp);
L
Linus Torvalds 已提交
1490 1491
	set_fs(oldfs);

1492
	if (host_err < 0)
L
Linus Torvalds 已提交
1493
		goto out_nfserr;
1494
	*lenp = host_err;
L
Linus Torvalds 已提交
1495 1496 1497 1498 1499
	err = 0;
out:
	return err;

out_nfserr:
1500
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
1508
__be32
L
Linus Torvalds 已提交
1509 1510 1511 1512 1513 1514 1515
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;
1516 1517
	__be32		err, cerr;
	int		host_err;
L
Linus Torvalds 已提交
1518 1519 1520 1521 1522 1523 1524 1525

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

M
Miklos Szeredi 已提交
1526
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1527 1528 1529 1530 1531
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
1532
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1533 1534 1535
	if (IS_ERR(dnew))
		goto out_nfserr;

1536 1537 1538 1539
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

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

1553
	if (!host_err) {
L
Linus Torvalds 已提交
1554
		if (EX_ISSYNC(fhp->fh_export))
1555 1556 1557
			host_err = nfsd_sync_dir(dentry);
	}
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1558 1559
	fh_unlock(fhp);

1560 1561
	mnt_drop_write(fhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1562 1563 1564 1565 1566 1567 1568
	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
1569
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1570 1571 1572 1573 1574 1575 1576
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1577
__be32
L
Linus Torvalds 已提交
1578 1579 1580 1581 1582
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;
1583 1584
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1585

M
Miklos Szeredi 已提交
1586
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1587 1588
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1589
	err = fh_verify(rqstp, tfhp, -S_IFDIR, NFSD_MAY_NOP);
L
Linus Torvalds 已提交
1590 1591 1592 1593 1594 1595 1596 1597 1598 1599
	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
1600
	fh_lock_nested(ffhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1601 1602 1603 1604
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
1605
	host_err = PTR_ERR(dnew);
L
Linus Torvalds 已提交
1606 1607 1608 1609 1610 1611
	if (IS_ERR(dnew))
		goto out_nfserr;

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

1612 1613 1614 1615 1616
	host_err = mnt_want_write(tfhp->fh_export->ex_path.mnt);
	if (host_err) {
		err = nfserrno(host_err);
		goto out_dput;
	}
1617 1618
	host_err = vfs_link(dold, dirp, dnew);
	if (!host_err) {
L
Linus Torvalds 已提交
1619
		if (EX_ISSYNC(ffhp->fh_export)) {
1620
			err = nfserrno(nfsd_sync_dir(ddir));
L
Linus Torvalds 已提交
1621 1622
			write_inode_now(dest, 1);
		}
1623
		err = 0;
L
Linus Torvalds 已提交
1624
	} else {
1625
		if (host_err == -EXDEV && rqstp->rq_vers == 2)
L
Linus Torvalds 已提交
1626 1627
			err = nfserr_acces;
		else
1628
			err = nfserrno(host_err);
L
Linus Torvalds 已提交
1629
	}
1630 1631
	mnt_drop_write(tfhp->fh_export->ex_path.mnt);
out_dput:
L
Linus Torvalds 已提交
1632
	dput(dnew);
1633 1634
out_unlock:
	fh_unlock(ffhp);
L
Linus Torvalds 已提交
1635 1636 1637 1638
out:
	return err;

out_nfserr:
1639
	err = nfserrno(host_err);
1640
	goto out_unlock;
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645 1646
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
1647
__be32
L
Linus Torvalds 已提交
1648 1649 1650 1651 1652
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;
1653 1654
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1655

M
Miklos Szeredi 已提交
1656
	err = fh_verify(rqstp, ffhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1657 1658
	if (err)
		goto out;
M
Miklos Szeredi 已提交
1659
	err = fh_verify(rqstp, tfhp, S_IFDIR, NFSD_MAY_CREATE);
L
Linus Torvalds 已提交
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669
	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;
1670
	if (ffhp->fh_export != tfhp->fh_export)
L
Linus Torvalds 已提交
1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684
		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);
1685
	host_err = PTR_ERR(odentry);
L
Linus Torvalds 已提交
1686 1687 1688
	if (IS_ERR(odentry))
		goto out_nfserr;

1689
	host_err = -ENOENT;
L
Linus Torvalds 已提交
1690 1691
	if (!odentry->d_inode)
		goto out_dput_old;
1692
	host_err = -EINVAL;
L
Linus Torvalds 已提交
1693 1694 1695 1696
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
1697
	host_err = PTR_ERR(ndentry);
L
Linus Torvalds 已提交
1698 1699
	if (IS_ERR(ndentry))
		goto out_dput_old;
1700
	host_err = -ENOTEMPTY;
L
Linus Torvalds 已提交
1701 1702 1703
	if (ndentry == trap)
		goto out_dput_new;

1704
	if (svc_msnfs(ffhp) &&
L
Linus Torvalds 已提交
1705 1706
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
1707
			host_err = -EPERM;
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
			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;

1718 1719 1720 1721 1722
	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
1723 1724
	}

1725 1726
	mnt_drop_write(ffhp->fh_export->ex_path.mnt);

L
Linus Torvalds 已提交
1727 1728 1729 1730 1731
 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
1732
	err = nfserrno(host_err);
L
Linus Torvalds 已提交
1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750

	/* 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
 */
1751
__be32
L
Linus Torvalds 已提交
1752 1753 1754 1755 1756
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
1757 1758
	__be32		err;
	int		host_err;
L
Linus Torvalds 已提交
1759 1760 1761 1762

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
M
Miklos Szeredi 已提交
1763
	err = fh_verify(rqstp, fhp, S_IFDIR, NFSD_MAY_REMOVE);
L
Linus Torvalds 已提交
1764 1765 1766
	if (err)
		goto out;

P
Peter Zijlstra 已提交
1767
	fh_lock_nested(fhp, I_MUTEX_PARENT);
L
Linus Torvalds 已提交
1768 1769 1770 1771
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
1772
	host_err = PTR_ERR(rdentry);
L
Linus Torvalds 已提交
1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
	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;

1785 1786 1787 1788
	host_err = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (host_err)
		goto out_nfserr;

L
Linus Torvalds 已提交
1789 1790 1791 1792
	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
1793
			host_err = -EPERM;
L
Linus Torvalds 已提交
1794 1795
		} else
#endif
1796
		host_err = vfs_unlink(dirp, rdentry);
L
Linus Torvalds 已提交
1797
	} else { /* It's RMDIR */
1798
		host_err = vfs_rmdir(dirp, rdentry);
L
Linus Torvalds 已提交
1799 1800 1801 1802
	}

	dput(rdentry);

1803
	if (host_err)
1804
		goto out_drop;
1805 1806
	if (EX_ISSYNC(fhp->fh_export))
		host_err = nfsd_sync_dir(dentry);
L
Linus Torvalds 已提交
1807

1808 1809
out_drop:
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
L
Linus Torvalds 已提交
1810
out_nfserr:
1811
	err = nfserrno(host_err);
1812 1813
out:
	return err;
L
Linus Torvalds 已提交
1814 1815 1816 1817 1818 1819
}

/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
1820
__be32
L
Linus Torvalds 已提交
1821
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
1822
	     struct readdir_cd *cdp, filldir_t func)
L
Linus Torvalds 已提交
1823
{
1824 1825
	__be32		err;
	int 		host_err;
L
Linus Torvalds 已提交
1826 1827 1828
	struct file	*file;
	loff_t		offset = *offsetp;

M
Miklos Szeredi 已提交
1829
	err = nfsd_open(rqstp, fhp, S_IFDIR, NFSD_MAY_READ, &file);
L
Linus Torvalds 已提交
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
	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 */
1847
		host_err = vfs_readdir(file, func, cdp);
1848 1849 1850
	} while (host_err >=0 && cdp->err == nfs_ok);
	if (host_err)
		err = nfserrno(host_err);
L
Linus Torvalds 已提交
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866
	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
 */
1867
__be32
1868
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat, int access)
L
Linus Torvalds 已提交
1869
{
1870
	__be32 err = fh_verify(rqstp, fhp, 0, NFSD_MAY_NOP | access);
1871
	if (!err && vfs_statfs(fhp->fh_dentry,stat))
L
Linus Torvalds 已提交
1872 1873 1874 1875
		err = nfserr_io;
	return err;
}

J
J. Bruce Fields 已提交
1876
static int exp_rdonly(struct svc_rqst *rqstp, struct svc_export *exp)
1877
{
J
J. Bruce Fields 已提交
1878
	return nfsexp_flags(rqstp, exp) & NFSEXP_READONLY;
1879 1880
}

L
Linus Torvalds 已提交
1881 1882 1883
/*
 * Check for a user's access permissions to this inode.
 */
1884
__be32
1885 1886
nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
					struct dentry *dentry, int acc)
L
Linus Torvalds 已提交
1887 1888 1889 1890
{
	struct inode	*inode = dentry->d_inode;
	int		err;

M
Miklos Szeredi 已提交
1891
	if (acc == NFSD_MAY_NOP)
L
Linus Torvalds 已提交
1892 1893 1894 1895
		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
1896 1897 1898 1899 1900 1901 1902
		(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
1903 1904 1905
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
1906
		__mnt_is_readonly(exp->ex_path.mnt)?	" ro" : "");
L
Linus Torvalds 已提交
1907 1908 1909 1910 1911 1912 1913 1914
	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
1915 1916
	if (!(acc & NFSD_MAY_LOCAL_ACCESS))
		if (acc & (NFSD_MAY_WRITE | NFSD_MAY_SATTR | NFSD_MAY_TRUNC)) {
1917 1918
			if (exp_rdonly(rqstp, exp) ||
			    __mnt_is_readonly(exp->ex_path.mnt))
L
Linus Torvalds 已提交
1919
				return nfserr_rofs;
M
Miklos Szeredi 已提交
1920
			if (/* (acc & NFSD_MAY_WRITE) && */ IS_IMMUTABLE(inode))
L
Linus Torvalds 已提交
1921 1922
				return nfserr_perm;
		}
M
Miklos Szeredi 已提交
1923
	if ((acc & NFSD_MAY_TRUNC) && IS_APPEND(inode))
L
Linus Torvalds 已提交
1924 1925
		return nfserr_perm;

M
Miklos Szeredi 已提交
1926
	if (acc & NFSD_MAY_LOCK) {
L
Linus Torvalds 已提交
1927 1928 1929 1930 1931 1932 1933
		/* 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
1934
			acc = NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE;
L
Linus Torvalds 已提交
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949
	}
	/*
	 * 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
1950
	if ((acc & NFSD_MAY_OWNER_OVERRIDE) &&
L
Linus Torvalds 已提交
1951 1952 1953
	    inode->i_uid == current->fsuid)
		return 0;

M
Miklos Szeredi 已提交
1954
	/* This assumes  NFSD_MAY_{READ,WRITE,EXEC} == MAY_{READ,WRITE,EXEC} */
1955
	err = inode_permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC));
L
Linus Torvalds 已提交
1956 1957 1958

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
M
Miklos Szeredi 已提交
1959
	    acc == (NFSD_MAY_READ | NFSD_MAY_OWNER_OVERRIDE))
1960
		err = inode_permission(inode, MAY_EXEC);
L
Linus Torvalds 已提交
1961 1962 1963 1964 1965 1966 1967

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
1968 1969 1970
	struct raparms *raparm, *last_raparm;
	unsigned int i;

L
Linus Torvalds 已提交
1971
	dprintk("nfsd: freeing readahead buffers.\n");
1972 1973 1974 1975 1976 1977 1978 1979 1980 1981

	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
1982 1983 1984 1985 1986 1987 1988 1989
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;
1990 1991
	int	j = 0;
	int	nperbucket;
1992
	struct raparms **raparm = NULL;
L
Linus Torvalds 已提交
1993

1994

1995
	if (raparm_hash[0].pb_head)
L
Linus Torvalds 已提交
1996
		return 0;
1997 1998 1999 2000
	nperbucket = DIV_ROUND_UP(cache_size, RAPARM_HASH_SIZE);
	if (nperbucket < 2)
		nperbucket = 2;
	cache_size = nperbucket * RAPARM_HASH_SIZE;
2001 2002

	dprintk("nfsd: allocating %d readahead buffers.\n", cache_size);
2003 2004

	for (i = 0; i < RAPARM_HASH_SIZE; i++) {
2005
		spin_lock_init(&raparm_hash[i].pb_lock);
2006 2007 2008 2009 2010 2011 2012 2013 2014

		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;
2015 2016
	}

L
Linus Torvalds 已提交
2017 2018
	nfsdstats.ra_size = cache_size;
	return 0;
2019 2020 2021 2022 2023

out_nomem:
	dprintk("nfsd: kmalloc failed, freeing readahead buffers\n");
	nfsd_racache_shutdown();
	return -ENOMEM;
L
Linus Torvalds 已提交
2024
}
2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035

#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;

2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046
	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:
2047 2048 2049
		return ERR_PTR(-EOPNOTSUPP);
	}

2050 2051 2052
	size = nfsd_getxattr(fhp->fh_dentry, name, &value);
	if (size < 0)
		return ERR_PTR(size);
2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072

	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:
2073
			name = POSIX_ACL_XATTR_ACCESS;
2074 2075
			break;
		case ACL_TYPE_DEFAULT:
2076
			name = POSIX_ACL_XATTR_DEFAULT;
2077 2078 2079 2080 2081 2082
			break;
		default:
			return -EOPNOTSUPP;
	}

	if (acl && acl->a_count) {
2083
		size = posix_acl_xattr_size(acl->a_count);
2084 2085 2086
		value = kmalloc(size, GFP_KERNEL);
		if (!value)
			return -ENOMEM;
2087 2088
		error = posix_acl_to_xattr(acl, value, size);
		if (error < 0)
2089
			goto getout;
2090
		size = error;
2091 2092 2093
	} else
		size = 0;

2094 2095 2096
	error = mnt_want_write(fhp->fh_export->ex_path.mnt);
	if (error)
		goto getout;
2097
	if (size)
2098
		error = vfs_setxattr(fhp->fh_dentry, name, value, size, 0);
2099 2100 2101 2102
	else {
		if (!S_ISDIR(inode->i_mode) && type == ACL_TYPE_DEFAULT)
			error = 0;
		else {
2103
			error = vfs_removexattr(fhp->fh_dentry, name);
2104 2105 2106 2107
			if (error == -ENODATA)
				error = 0;
		}
	}
2108
	mnt_drop_write(fhp->fh_export->ex_path.mnt);
2109 2110 2111 2112 2113 2114

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