nfs4state.c 132.6 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 26 27 28 29 30 31 32 33 34
/*
*  Copyright (c) 2001 The Regents of the University of Michigan.
*  All rights reserved.
*
*  Kendrick Smith <kmsmith@umich.edu>
*  Andy Adamson <kandros@umich.edu>
*
*  Redistribution and use in source and binary forms, with or without
*  modification, are permitted provided that the following conditions
*  are met:
*
*  1. Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*  2. Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in the
*     documentation and/or other materials provided with the distribution.
*  3. Neither the name of the University nor the names of its
*     contributors may be used to endorse or promote products derived
*     from this software without specific prior written permission.
*
*  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
*  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
*  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
*  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
*  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
*  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
*  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
*  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
*  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
*  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
*  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/

35
#include <linux/file.h>
36
#include <linux/fs.h>
37
#include <linux/slab.h>
38
#include <linux/namei.h>
39
#include <linux/swap.h>
40
#include <linux/pagemap.h>
41
#include <linux/ratelimit.h>
42
#include <linux/sunrpc/svcauth_gss.h>
43
#include <linux/sunrpc/addr.h>
44
#include "xdr4.h"
45
#include "vfs.h"
46
#include "current_stateid.h"
L
Linus Torvalds 已提交
47

48 49
#include "netns.h"

L
Linus Torvalds 已提交
50 51
#define NFSDDBG_FACILITY                NFSDDBG_PROC

52 53 54 55 56 57 58 59
#define all_ones {{~0,~0},~0}
static const stateid_t one_stateid = {
	.si_generation = ~0,
	.si_opaque = all_ones,
};
static const stateid_t zero_stateid = {
	/* all fields zero */
};
60 61 62
static const stateid_t currentstateid = {
	.si_generation = 1,
};
63

A
Andy Adamson 已提交
64
static u64 current_sessionid = 1;
65

66 67
#define ZERO_STATEID(stateid) (!memcmp((stateid), &zero_stateid, sizeof(stateid_t)))
#define ONE_STATEID(stateid)  (!memcmp((stateid), &one_stateid, sizeof(stateid_t)))
68
#define CURRENT_STATEID(stateid) (!memcmp((stateid), &currentstateid, sizeof(stateid_t)))
L
Linus Torvalds 已提交
69 70

/* forward declarations */
71
static int check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner);
L
Linus Torvalds 已提交
72

73 74 75
/* Locking: */

/* Currently used for almost all code touching nfsv4 state: */
I
Ingo Molnar 已提交
76
static DEFINE_MUTEX(client_mutex);
L
Linus Torvalds 已提交
77

78 79 80 81 82 83 84
/*
 * Currently used for the del_recall_lru and file hash table.  In an
 * effort to decrease the scope of the client_mutex, this spinlock may
 * eventually cover more:
 */
static DEFINE_SPINLOCK(recall_lock);

85 86
static struct kmem_cache *openowner_slab = NULL;
static struct kmem_cache *lockowner_slab = NULL;
87 88 89
static struct kmem_cache *file_slab = NULL;
static struct kmem_cache *stateid_slab = NULL;
static struct kmem_cache *deleg_slab = NULL;
N
NeilBrown 已提交
90

L
Linus Torvalds 已提交
91 92 93
void
nfs4_lock_state(void)
{
I
Ingo Molnar 已提交
94
	mutex_lock(&client_mutex);
L
Linus Torvalds 已提交
95 96
}

97
static void free_session(struct nfsd4_session *);
98

99 100 101 102 103 104
void nfsd4_put_session(struct nfsd4_session *ses)
{
	atomic_dec(&ses->se_ref);
}

static bool is_session_dead(struct nfsd4_session *ses)
105
{
106 107 108 109 110 111 112 113 114
	return ses->se_flags & NFS4_SESSION_DEAD;
}

static __be32 mark_session_dead_locked(struct nfsd4_session *ses)
{
	if (atomic_read(&ses->se_ref))
		return nfserr_jukebox;
	ses->se_flags |= NFS4_SESSION_DEAD;
	return nfs_ok;
115 116
}

117
static __be32 nfsd4_get_session_locked(struct nfsd4_session *ses)
118
{
119 120 121 122
	if (is_session_dead(ses))
		return nfserr_badsession;
	atomic_inc(&ses->se_ref);
	return nfs_ok;
123 124
}

L
Linus Torvalds 已提交
125 126 127
void
nfs4_unlock_state(void)
{
I
Ingo Molnar 已提交
128
	mutex_unlock(&client_mutex);
L
Linus Torvalds 已提交
129 130
}

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
static bool is_client_expired(struct nfs4_client *clp)
{
	return clp->cl_time == 0;
}

static __be32 mark_client_expired_locked(struct nfs4_client *clp)
{
	if (atomic_read(&clp->cl_refcount))
		return nfserr_jukebox;
	clp->cl_time = 0;
	return nfs_ok;
}

static __be32 mark_client_expired(struct nfs4_client *clp)
{
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
	__be32 ret;

	spin_lock(&nn->client_lock);
	ret = mark_client_expired_locked(clp);
	spin_unlock(&nn->client_lock);
	return ret;
}

static __be32 get_client_locked(struct nfs4_client *clp)
{
	if (is_client_expired(clp))
		return nfserr_expired;
	atomic_inc(&clp->cl_refcount);
	return nfs_ok;
}

/* must be called under the client_lock */
static inline void
renew_client_locked(struct nfs4_client *clp)
{
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	if (is_client_expired(clp)) {
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

	dprintk("renewing client (clientid %08x/%08x)\n",
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
	list_move_tail(&clp->cl_lru, &nn->client_lru);
	clp->cl_time = get_seconds();
}

static inline void
renew_client(struct nfs4_client *clp)
{
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
	renew_client_locked(clp);
	spin_unlock(&nn->client_lock);
}

void put_client_renew_locked(struct nfs4_client *clp)
{
	if (!atomic_dec_and_test(&clp->cl_refcount))
		return;
	if (!is_client_expired(clp))
		renew_client_locked(clp);
}

void put_client_renew(struct nfs4_client *clp)
{
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
		return;
	if (!is_client_expired(clp))
		renew_client_locked(clp);
	spin_unlock(&nn->client_lock);
}


L
Linus Torvalds 已提交
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
static inline u32
opaque_hashval(const void *ptr, int nbytes)
{
	unsigned char *cptr = (unsigned char *) ptr;

	u32 x = 0;
	while (nbytes--) {
		x *= 37;
		x += *cptr++;
	}
	return x;
}

static struct list_head del_recall_lru;

230 231 232 233 234
static void nfsd4_free_file(struct nfs4_file *f)
{
	kmem_cache_free(file_slab, f);
}

235 236 237
static inline void
put_nfs4_file(struct nfs4_file *fi)
{
238
	if (atomic_dec_and_lock(&fi->fi_ref, &recall_lock)) {
239
		hlist_del(&fi->fi_hash);
240 241
		spin_unlock(&recall_lock);
		iput(fi->fi_inode);
242
		nfsd4_free_file(fi);
243
	}
244 245 246 247 248
}

static inline void
get_nfs4_file(struct nfs4_file *fi)
{
249
	atomic_inc(&fi->fi_ref);
250 251
}

252
static int num_delegations;
253
unsigned long max_delegations;
254 255 256 257 258

/*
 * Open owner state (share locks)
 */

259 260 261 262
/* hash tables for lock and open owners */
#define OWNER_HASH_BITS              8
#define OWNER_HASH_SIZE             (1 << OWNER_HASH_BITS)
#define OWNER_HASH_MASK             (OWNER_HASH_SIZE - 1)
263

264
static unsigned int ownerstr_hashval(u32 clientid, struct xdr_netobj *ownername)
265 266 267 268 269
{
	unsigned int ret;

	ret = opaque_hashval(ownername->data, ownername->len);
	ret += clientid;
270
	return ret & OWNER_HASH_MASK;
271
}
272 273 274 275

/* hash table for nfs4_file */
#define FILE_HASH_BITS                   8
#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
S
Shan Wei 已提交
276

277 278 279 280 281 282
static unsigned int file_hashval(struct inode *ino)
{
	/* XXX: why are we hashing on inode pointer, anyway? */
	return hash_ptr(ino, FILE_HASH_BITS);
}

283
static struct hlist_head file_hashtbl[FILE_HASH_SIZE];
284

285
static void __nfs4_file_get_access(struct nfs4_file *fp, int oflag)
286
{
287
	WARN_ON_ONCE(!(fp->fi_fds[oflag] || fp->fi_fds[O_RDWR]));
288 289 290
	atomic_inc(&fp->fi_access[oflag]);
}

291 292 293 294 295 296 297 298 299 300
static void nfs4_file_get_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_get_access(fp, O_RDONLY);
		__nfs4_file_get_access(fp, O_WRONLY);
	} else
		__nfs4_file_get_access(fp, oflag);
}

static void nfs4_file_put_fd(struct nfs4_file *fp, int oflag)
301 302 303 304 305 306 307
{
	if (fp->fi_fds[oflag]) {
		fput(fp->fi_fds[oflag]);
		fp->fi_fds[oflag] = NULL;
	}
}

308
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
309 310 311
{
	if (atomic_dec_and_test(&fp->fi_access[oflag])) {
		nfs4_file_put_fd(fp, oflag);
312 313 314 315 316 317 318 319
		/*
		 * It's also safe to get rid of the RDWR open *if*
		 * we no longer have need of the other kind of access
		 * or if we already have the other kind of open:
		 */
		if (fp->fi_fds[1-oflag]
			|| atomic_read(&fp->fi_access[1 - oflag]) == 0)
			nfs4_file_put_fd(fp, O_RDWR);
320 321 322
	}
}

323 324 325 326 327 328 329 330 331
static void nfs4_file_put_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_put_access(fp, O_RDONLY);
		__nfs4_file_put_access(fp, O_WRONLY);
	} else
		__nfs4_file_put_access(fp, oflag);
}

J
J. Bruce Fields 已提交
332 333
static struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl, struct
kmem_cache *slab)
334
{
J
J. Bruce Fields 已提交
335 336 337
	struct idr *stateids = &cl->cl_stateids;
	static int min_stateid = 0;
	struct nfs4_stid *stid;
J
J. Bruce Fields 已提交
338
	int new_id;
339

J
J. Bruce Fields 已提交
340 341 342 343
	stid = kmem_cache_alloc(slab, GFP_KERNEL);
	if (!stid)
		return NULL;

T
Tejun Heo 已提交
344 345
	new_id = idr_alloc(stateids, stid, min_stateid, 0, GFP_KERNEL);
	if (new_id < 0)
J
J. Bruce Fields 已提交
346
		goto out_free;
347
	stid->sc_client = cl;
J
J. Bruce Fields 已提交
348 349 350
	stid->sc_type = 0;
	stid->sc_stateid.si_opaque.so_id = new_id;
	stid->sc_stateid.si_opaque.so_clid = cl->cl_clientid;
351
	/* Will be incremented before return to client: */
J
J. Bruce Fields 已提交
352
	stid->sc_stateid.si_generation = 0;
353 354

	/*
J
J. Bruce Fields 已提交
355 356 357 358 359 360 361
	 * It shouldn't be a problem to reuse an opaque stateid value.
	 * I don't think it is for 4.1.  But with 4.0 I worry that, for
	 * example, a stray write retransmission could be accepted by
	 * the server when it should have been rejected.  Therefore,
	 * adopt a trick from the sctp code to attempt to maximize the
	 * amount of time until an id is reused, by ensuring they always
	 * "increase" (mod INT_MAX):
362
	 */
J
J. Bruce Fields 已提交
363 364 365 366 367 368 369 370

	min_stateid = new_id+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return stid;
out_free:
	kfree(stid);
	return NULL;
371 372
}

373 374 375 376 377
static struct nfs4_ol_stateid * nfs4_alloc_stateid(struct nfs4_client *clp)
{
	return openlockstateid(nfs4_alloc_stid(clp, stateid_slab));
}

L
Linus Torvalds 已提交
378
static struct nfs4_delegation *
379
alloc_init_deleg(struct nfs4_client *clp, struct nfs4_ol_stateid *stp, struct svc_fh *current_fh, u32 type)
L
Linus Torvalds 已提交
380 381 382 383 384
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
385 386 387 388 389 390 391
	/*
	 * Major work on the lease subsystem (for example, to support
	 * calbacks on stat) will be required before we can support
	 * write delegations properly.
	 */
	if (type != NFS4_OPEN_DELEGATE_READ)
		return NULL;
392 393
	if (fp->fi_had_conflict)
		return NULL;
394
	if (num_delegations > max_delegations)
395
		return NULL;
396
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
397
	if (dp == NULL)
L
Linus Torvalds 已提交
398
		return dp;
J
J. Bruce Fields 已提交
399
	dp->dl_stid.sc_type = NFS4_DELEG_STID;
400 401
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
402 403
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
404 405
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
406
	num_delegations++;
407 408
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
409
	INIT_LIST_HEAD(&dp->dl_recall_lru);
410
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
411 412
	dp->dl_file = fp;
	dp->dl_type = type;
413
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
414 415
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
416
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
417 418 419
	return dp;
}

F
Fengguang Wu 已提交
420
static void free_stid(struct nfs4_stid *s, struct kmem_cache *slab)
J
J. Bruce Fields 已提交
421 422 423 424 425 426 427
{
	struct idr *stateids = &s->sc_client->cl_stateids;

	idr_remove(stateids, s->sc_stateid.si_opaque.so_id);
	kmem_cache_free(slab, s);
}

L
Linus Torvalds 已提交
428 429 430 431 432
void
nfs4_put_delegation(struct nfs4_delegation *dp)
{
	if (atomic_dec_and_test(&dp->dl_count)) {
		dprintk("NFSD: freeing dp %p\n",dp);
433
		put_nfs4_file(dp->dl_file);
J
J. Bruce Fields 已提交
434
		free_stid(&dp->dl_stid, deleg_slab);
435
		num_delegations--;
L
Linus Torvalds 已提交
436 437 438
	}
}

439
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
440
{
441 442 443
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
444
		fput(fp->fi_deleg_file);
445 446
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
447 448
}

J
J. Bruce Fields 已提交
449 450
static void unhash_stid(struct nfs4_stid *s)
{
J
J. Bruce Fields 已提交
451
	s->sc_type = 0;
J
J. Bruce Fields 已提交
452 453
}

L
Linus Torvalds 已提交
454 455 456 457
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
458
	unhash_stid(&dp->dl_stid);
459
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
460
	spin_lock(&recall_lock);
461
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
462 463
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
464
	nfs4_put_deleg_lease(dp->dl_file);
L
Linus Torvalds 已提交
465 466 467 468 469 470 471
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

472 473 474 475 476 477 478 479 480 481
static unsigned int clientid_hashval(u32 id)
{
	return id & CLIENT_HASH_MASK;
}

static unsigned int clientstr_hashval(const char *name)
{
	return opaque_hashval(name, 8) & CLIENT_HASH_MASK;
}

482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499
/*
 * We store the NONE, READ, WRITE, and BOTH bits separately in the
 * st_{access,deny}_bmap field of the stateid, in order to track not
 * only what share bits are currently in force, but also what
 * combinations of share bits previous opens have used.  This allows us
 * to enforce the recommendation of rfc 3530 14.2.19 that the server
 * return an error if the client attempt to downgrade to a combination
 * of share bits not explicable by closing some of its previous opens.
 *
 * XXX: This enforcement is actually incomplete, since we don't keep
 * track of access/deny bit combinations; so, e.g., we allow:
 *
 *	OPEN allow read, deny write
 *	OPEN allow both, deny none
 *	DOWNGRADE allow read, deny none
 *
 * which we should reject.
 */
500 501
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
502
	int i;
503
	unsigned int access = 0;
504 505 506

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
507
			access |= i;
508
	}
509
	return access;
510 511
}

512
static bool
513
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
514 515
	unsigned int access, deny;

516 517
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
518
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
519 520
		return false;
	return true;
521 522
}

523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
/* set share access for a given stateid */
static inline void
set_access(u32 access, struct nfs4_ol_stateid *stp)
{
	__set_bit(access, &stp->st_access_bmap);
}

/* clear share access for a given stateid */
static inline void
clear_access(u32 access, struct nfs4_ol_stateid *stp)
{
	__clear_bit(access, &stp->st_access_bmap);
}

/* test whether a given stateid has access */
static inline bool
test_access(u32 access, struct nfs4_ol_stateid *stp)
{
	return test_bit(access, &stp->st_access_bmap);
}

544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562
/* set share deny for a given stateid */
static inline void
set_deny(u32 access, struct nfs4_ol_stateid *stp)
{
	__set_bit(access, &stp->st_deny_bmap);
}

/* clear share deny for a given stateid */
static inline void
clear_deny(u32 access, struct nfs4_ol_stateid *stp)
{
	__clear_bit(access, &stp->st_deny_bmap);
}

/* test whether a given stateid is denying specific access */
static inline bool
test_deny(u32 access, struct nfs4_ol_stateid *stp)
{
	return test_bit(access, &stp->st_deny_bmap);
563 564 565 566
}

static int nfs4_access_to_omode(u32 access)
{
567
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
568 569 570 571 572 573 574
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
575 576
	WARN_ON_ONCE(1);
	return O_RDONLY;
577 578
}

579 580 581 582 583 584 585 586 587 588 589 590 591 592
/* release all access and file references for a given stateid */
static void
release_all_access(struct nfs4_ol_stateid *stp)
{
	int i;

	for (i = 1; i < 4; i++) {
		if (test_access(i, stp))
			nfs4_file_put_access(stp->st_file,
					     nfs4_access_to_omode(i));
		clear_access(i, stp);
	}
}

593
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
594 595 596 597 598
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

599
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
600
{
601
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
602
	put_nfs4_file(stp->st_file);
603 604 605
	stp->st_file = NULL;
}

606
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
607
{
J
J. Bruce Fields 已提交
608
	free_stid(&stp->st_stid, stateid_slab);
609 610
}

611
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
612 613 614 615
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
616
	unhash_stid(&stp->st_stid);
617 618
	file = find_any_file(stp->st_file);
	if (file)
619
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
620
	close_generic_stateid(stp);
621 622 623
	free_generic_stateid(stp);
}

624
static void unhash_lockowner(struct nfs4_lockowner *lo)
625
{
626
	struct nfs4_ol_stateid *stp;
627

628 629
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
630
	list_del(&lo->lo_owner_ino_hash);
631 632
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
633
				struct nfs4_ol_stateid, st_perstateowner);
634 635 636 637
		release_lock_stateid(stp);
	}
}

638
static void release_lockowner(struct nfs4_lockowner *lo)
639
{
640 641
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
642 643 644
}

static void
645
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
646
{
647
	struct nfs4_lockowner *lo;
648 649

	while (!list_empty(&open_stp->st_lockowners)) {
650 651 652
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
653 654 655
	}
}

656
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
657 658 659
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
660 661 662 663 664 665
	close_generic_stateid(stp);
}

static void release_open_stateid(struct nfs4_ol_stateid *stp)
{
	unhash_open_stateid(stp);
J
J. Bruce Fields 已提交
666
	unhash_stid(&stp->st_stid);
667 668 669
	free_generic_stateid(stp);
}

670
static void unhash_openowner(struct nfs4_openowner *oo)
671
{
672
	struct nfs4_ol_stateid *stp;
673

674 675 676 677
	list_del(&oo->oo_owner.so_strhash);
	list_del(&oo->oo_perclient);
	while (!list_empty(&oo->oo_owner.so_stateids)) {
		stp = list_first_entry(&oo->oo_owner.so_stateids,
678
				struct nfs4_ol_stateid, st_perstateowner);
679
		release_open_stateid(stp);
680 681 682
	}
}

683 684 685 686 687
static void release_last_closed_stateid(struct nfs4_openowner *oo)
{
	struct nfs4_ol_stateid *s = oo->oo_last_closed_stid;

	if (s) {
J
J. Bruce Fields 已提交
688
		unhash_stid(&s->st_stid);
689 690 691 692 693
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

694
static void release_openowner(struct nfs4_openowner *oo)
695
{
696 697
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
698
	release_last_closed_stateid(oo);
699
	nfs4_free_openowner(oo);
700 701
}

M
Marc Eshel 已提交
702 703 704 705 706 707 708 709
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

710
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
711 712 713 714 715 716
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
	u32 *ptr = (u32 *)(&sessionid->data[0]);
	dprintk("%s: %u:%u:%u:%u\n", fn, ptr[0], ptr[1], ptr[2], ptr[3]);
}
717 718 719 720 721 722 723
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

M
Marc Eshel 已提交
724

A
Andy Adamson 已提交
725 726 727 728 729 730 731 732 733 734 735 736 737
static void
gen_sessionid(struct nfsd4_session *ses)
{
	struct nfs4_client *clp = ses->se_client;
	struct nfsd4_sessionid *sid;

	sid = (struct nfsd4_sessionid *)ses->se_sessionid.data;
	sid->clientid = clp->cl_clientid;
	sid->sequence = current_sessionid++;
	sid->reserved = 0;
}

/*
738 739 740 741 742 743 744 745 746 747 748 749 750
 * The protocol defines ca_maxresponssize_cached to include the size of
 * the rpc header, but all we need to cache is the data starting after
 * the end of the initial SEQUENCE operation--the rest we regenerate
 * each time.  Therefore we can advertise a ca_maxresponssize_cached
 * value that is the number of bytes in our cache plus a few additional
 * bytes.  In order to stay on the safe side, and not promise more than
 * we can cache, those additional bytes must be the minimum possible: 24
 * bytes of rpc header (xid through accept state, with AUTH_NULL
 * verifier), 12 for the compound header (with zero-length tag), and 44
 * for the SEQUENCE op response:
 */
#define NFSD_MIN_HDR_SEQ_SZ  (24 + 12 + 44)

751 752 753 754 755 756 757 758 759
static void
free_session_slots(struct nfsd4_session *ses)
{
	int i;

	for (i = 0; i < ses->se_fchannel.maxreqs; i++)
		kfree(ses->se_slots[i]);
}

760
/*
761 762 763 764 765 766 767 768
 * We don't actually need to cache the rpc and session headers, so we
 * can allocate a little less for each slot:
 */
static inline int slot_bytes(struct nfsd4_channel_attrs *ca)
{
	return ca->maxresp_cached - NFSD_MIN_HDR_SEQ_SZ;
}

769
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
770
{
771 772
	size -= NFSD_MIN_HDR_SEQ_SZ; /* We don't cache the rpc header */
	size = min_t(u32, size, NFSD_SLOT_CACHE_SIZE);
A
Andy Adamson 已提交
773

774 775
	return size;
}
A
Andy Adamson 已提交
776

777 778
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
779
 * re-negotiate active sessions and reduce their slot usage to make
780
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
781
 */
782
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
783
{
784
	int avail;
A
Andy Adamson 已提交
785

786
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
787

788
	spin_lock(&nfsd_drc_lock);
789 790
	avail = min((unsigned long)NFSD_MAX_MEM_PER_SESSION,
		    nfsd_drc_max_mem - nfsd_drc_mem_used);
791 792 793
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
794

795 796
	return num;
}
A
Andy Adamson 已提交
797

798 799
static void nfsd4_put_drc_mem(int slotsize, int num)
{
800
	spin_lock(&nfsd_drc_lock);
801
	nfsd_drc_mem_used -= slotsize * num;
802
	spin_unlock(&nfsd_drc_lock);
803
}
A
Andy Adamson 已提交
804

805
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
806 807 808
{
	struct nfsd4_session *new;
	int mem, i;
809

810 811 812
	BUILD_BUG_ON(NFSD_MAX_SLOTS_PER_SESSION * sizeof(struct nfsd4_slot *)
			+ sizeof(struct nfsd4_session) > PAGE_SIZE);
	mem = numslots * sizeof(struct nfsd4_slot *);
A
Andy Adamson 已提交
813

814 815 816
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
817
	/* allocate each struct nfsd4_slot and data cache in one piece */
818 819 820 821
	for (i = 0; i < numslots; i++) {
		mem = sizeof(struct nfsd4_slot) + slotsize;
		new->se_slots[i] = kzalloc(mem, GFP_KERNEL);
		if (!new->se_slots[i])
822 823
			goto out_free;
	}
824 825 826 827 828 829
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
830 831
}

832 833 834 835
static void init_forechannel_attrs(struct nfsd4_channel_attrs *new,
				   struct nfsd4_channel_attrs *req,
				   int numslots, int slotsize,
				   struct nfsd_net *nn)
A
Andy Adamson 已提交
836
{
837
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
838

839
	new->maxreqs = numslots;
840 841
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
842 843 844 845
	new->maxreq_sz = min_t(u32, req->maxreq_sz, maxrpc);
	new->maxresp_sz = min_t(u32, req->maxresp_sz, maxrpc);
	new->maxops = min_t(u32, req->maxops, NFSD_MAX_OPS_PER_COMPOUND);
}
A
Andy Adamson 已提交
846

847 848 849 850 851
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
852

853 854 855 856
static void nfsd4_conn_lost(struct svc_xpt_user *u)
{
	struct nfsd4_conn *c = container_of(u, struct nfsd4_conn, cn_xpt_user);
	struct nfs4_client *clp = c->cn_session->se_client;
A
Andy Adamson 已提交
857

858 859 860 861 862
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
863
	nfsd4_probe_callback(clp);
864
	spin_unlock(&clp->cl_lock);
865
}
A
Andy Adamson 已提交
866

867
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
868 869
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
870

871 872
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
873
		return NULL;
874 875
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
876
	conn->cn_flags = flags;
877 878 879
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
880

881 882 883 884
static void __nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
{
	conn->cn_session = ses;
	list_add(&conn->cn_persession, &ses->se_conns);
A
Andy Adamson 已提交
885 886
}

887
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
888
{
889
	struct nfs4_client *clp = ses->se_client;
890

891
	spin_lock(&clp->cl_lock);
892
	__nfsd4_hash_conn(conn, ses);
893
	spin_unlock(&clp->cl_lock);
894 895
}

896
static int nfsd4_register_conn(struct nfsd4_conn *conn)
897
{
898
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
899
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
900 901
}

902
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
903
{
904
	int ret;
A
Andy Adamson 已提交
905

906
	nfsd4_hash_conn(conn, ses);
907 908 909 910
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
911
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
912 913 914
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
915
}
A
Andy Adamson 已提交
916

917
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
918 919 920
{
	u32 dir = NFS4_CDFC4_FORE;

921
	if (cses->flags & SESSION4_BACK_CHAN)
922
		dir |= NFS4_CDFC4_BACK;
923
	return alloc_conn(rqstp, dir);
924 925 926
}

/* must be called under client_lock */
927
static void nfsd4_del_conns(struct nfsd4_session *s)
928
{
929 930
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
931

932 933 934 935 936
	spin_lock(&clp->cl_lock);
	while (!list_empty(&s->se_conns)) {
		c = list_first_entry(&s->se_conns, struct nfsd4_conn, cn_persession);
		list_del_init(&c->cn_persession);
		spin_unlock(&clp->cl_lock);
937

938 939
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
940

941 942 943
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
944
}
A
Andy Adamson 已提交
945

946 947 948 949 950 951 952
static void __free_session(struct nfsd4_session *ses)
{
	nfsd4_put_drc_mem(slot_bytes(&ses->se_fchannel), ses->se_fchannel.maxreqs);
	free_session_slots(ses);
	kfree(ses);
}

953
static void free_session(struct nfsd4_session *ses)
954
{
955
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);
956 957

	lockdep_assert_held(&nn->client_lock);
958
	nfsd4_del_conns(ses);
959
	__free_session(ses);
960 961
}

962 963
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan,
					   struct nfsd_net *nn)
964 965 966 967 968 969 970 971 972 973 974 975
{
	struct nfsd4_session *new;
	int numslots, slotsize;
	/*
	 * Note decreasing slot size below client's request may
	 * make it difficult for client to function correctly, whereas
	 * decreasing the number of slots will (just?) affect
	 * performance.  When short on memory we therefore prefer to
	 * decrease number of slots instead of their size.
	 */
	slotsize = nfsd4_sanitize_slot_size(fchan->maxresp_cached);
	numslots = nfsd4_get_drc_mem(slotsize, fchan->maxreqs);
976 977
	if (numslots < 1)
		return NULL;
978

979
	new = __alloc_session(slotsize, numslots);
980
	if (!new) {
981
		nfsd4_put_drc_mem(slotsize, numslots);
982
		return NULL;
983
	}
984
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize, nn);
985 986
	return new;
}
987

988
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
989 990
{
	int idx;
991
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
992

A
Andy Adamson 已提交
993 994 995
	new->se_client = clp;
	gen_sessionid(new);

996 997
	INIT_LIST_HEAD(&new->se_conns);

998
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
999
	new->se_flags = cses->flags;
1000
	new->se_cb_prog = cses->callback_prog;
1001
	new->se_cb_sec = cses->cb_sec;
1002
	atomic_set(&new->se_ref, 0);
1003
	idx = hash_sessionid(&new->se_sessionid);
1004
	spin_lock(&nn->client_lock);
1005
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
1006
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
1007
	list_add(&new->se_perclnt, &clp->cl_sessions);
1008
	spin_unlock(&clp->cl_lock);
1009
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
1010

1011
	if (cses->flags & SESSION4_BACK_CHAN) {
1012
		struct sockaddr *sa = svc_addr(rqstp);
1013 1014 1015 1016 1017 1018 1019
		/*
		 * This is a little silly; with sessions there's no real
		 * use for the callback address.  Use the peer address
		 * as a reasonable default for now, but consider fixing
		 * the rpc client not to require an address in the
		 * future:
		 */
1020 1021 1022
		rpc_copy_addr((struct sockaddr *)&clp->cl_cb_conn.cb_addr, sa);
		clp->cl_cb_conn.cb_addrlen = svc_addr_len(sa);
	}
A
Andy Adamson 已提交
1023 1024
}

1025
/* caller must hold client_lock */
M
Marc Eshel 已提交
1026
static struct nfsd4_session *
1027
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
1028 1029 1030
{
	struct nfsd4_session *elem;
	int idx;
1031
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
1032 1033 1034 1035

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
1036
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

	dprintk("%s: session not found\n", __func__);
	return NULL;
}

1047
/* caller must hold client_lock */
A
Andy Adamson 已提交
1048
static void
M
Marc Eshel 已提交
1049
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
1050 1051
{
	list_del(&ses->se_hash);
1052
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
1053
	list_del(&ses->se_perclnt);
1054
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
1055 1056
}

L
Linus Torvalds 已提交
1057 1058
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1059
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1060
{
1061
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1062
		return 0;
A
Andy Adamson 已提交
1063
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1064
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1065 1066 1067 1068 1069 1070 1071 1072
	return 1;
}

/* 
 * XXX Should we use a slab cache ?
 * This type of memory management is somewhat inefficient, but we use it
 * anyway since SETCLIENTID is not a common operation.
 */
1073
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1074 1075 1076
{
	struct nfs4_client *clp;

1077 1078 1079
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1080
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1081 1082 1083
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1084
	}
1085
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1086 1087 1088 1089 1090 1091
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1092
	struct nfsd_net __maybe_unused *nn = net_generic(clp->net, nfsd_net_id);
1093 1094

	lockdep_assert_held(&nn->client_lock);
1095 1096 1097 1098 1099
	while (!list_empty(&clp->cl_sessions)) {
		struct nfsd4_session *ses;
		ses = list_entry(clp->cl_sessions.next, struct nfsd4_session,
				se_perclnt);
		list_del(&ses->se_perclnt);
1100 1101
		WARN_ON_ONCE(atomic_read(&ses->se_ref));
		free_session(ses);
1102
	}
1103
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1104
	kfree(clp->cl_name.data);
M
majianpeng 已提交
1105
	idr_destroy(&clp->cl_stateids);
L
Linus Torvalds 已提交
1106 1107 1108
	kfree(clp);
}

B
Benny Halevy 已提交
1109 1110 1111 1112
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1113 1114
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1115
	list_del(&clp->cl_lru);
1116
	spin_lock(&clp->cl_lock);
1117 1118
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1119
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1120 1121
}

L
Linus Torvalds 已提交
1122
static void
1123
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1124
{
1125
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1126 1127
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1128
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1129 1130 1131

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1132 1133 1134
	while (!list_empty(&clp->cl_delegations)) {
		dp = list_entry(clp->cl_delegations.next, struct nfs4_delegation, dl_perclnt);
		list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140 1141
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	while (!list_empty(&reaplist)) {
		dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}
1142
	while (!list_empty(&clp->cl_openowners)) {
1143 1144
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1145
	}
1146
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1147 1148
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1149
	list_del(&clp->cl_idhash);
1150
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1151
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1152
	else
1153
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1154
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1155
	unhash_client_locked(clp);
1156 1157
	WARN_ON_ONCE(atomic_read(&clp->cl_refcount));
	free_client(clp);
1158
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1159 1160
}

1161 1162 1163 1164 1165 1166
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1167 1168 1169 1170
static void copy_verf(struct nfs4_client *target, nfs4_verifier *source)
{
	memcpy(target->cl_verifier.data, source->data,
			sizeof(target->cl_verifier.data));
L
Linus Torvalds 已提交
1171 1172
}

1173 1174
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1175 1176 1177 1178
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1179
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1180
{
1181 1182 1183 1184 1185 1186 1187
	if (source->cr_principal) {
		target->cr_principal =
				kstrdup(source->cr_principal, GFP_KERNEL);
		if (target->cr_principal == NULL)
			return -ENOMEM;
	} else
		target->cr_principal = NULL;
1188
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1189 1190 1191 1192
	target->cr_uid = source->cr_uid;
	target->cr_gid = source->cr_gid;
	target->cr_group_info = source->cr_group_info;
	get_group_info(target->cr_group_info);
1193
	return 0;
L
Linus Torvalds 已提交
1194 1195
}

1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
static long long
compare_blob(const struct xdr_netobj *o1, const struct xdr_netobj *o2)
{
	long long res;

	res = o1->len - o2->len;
	if (res)
		return res;
	return (long long)memcmp(o1->data, o2->data, o1->len);
}

1207
static int same_name(const char *n1, const char *n2)
1208
{
N
NeilBrown 已提交
1209
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1210 1211 1212
}

static int
1213 1214 1215
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1216 1217 1218
}

static int
1219 1220 1221
same_clid(clientid_t *cl1, clientid_t *cl2)
{
	return (cl1->cl_boot == cl2->cl_boot) && (cl1->cl_id == cl2->cl_id);
L
Linus Torvalds 已提交
1222 1223
}

1224 1225 1226 1227 1228 1229 1230
static bool groups_equal(struct group_info *g1, struct group_info *g2)
{
	int i;

	if (g1->ngroups != g2->ngroups)
		return false;
	for (i=0; i<g1->ngroups; i++)
1231
		if (!gid_eq(GROUP_AT(g1, i), GROUP_AT(g2, i)))
1232 1233 1234 1235
			return false;
	return true;
}

1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
/*
 * RFC 3530 language requires clid_inuse be returned when the
 * "principal" associated with a requests differs from that previously
 * used.  We use uid, gid's, and gss principal string as our best
 * approximation.  We also don't want to allow non-gss use of a client
 * established using gss: in theory cr_principal should catch that
 * change, but in practice cr_principal can be null even in the gss case
 * since gssd doesn't always pass down a principal string.
 */
static bool is_gss_cred(struct svc_cred *cr)
{
	/* Is cr_flavor one of the gss "pseudoflavors"?: */
	return (cr->cr_flavor > RPC_AUTH_MAXFLAVOR);
}


1252
static bool
1253 1254
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1255
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1256 1257
		|| (!uid_eq(cr1->cr_uid, cr2->cr_uid))
		|| (!gid_eq(cr1->cr_gid, cr2->cr_gid))
1258 1259 1260 1261 1262 1263
		|| !groups_equal(cr1->cr_group_info, cr2->cr_group_info))
		return false;
	if (cr1->cr_principal == cr2->cr_principal)
		return true;
	if (!cr1->cr_principal || !cr2->cr_principal)
		return false;
1264
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1265 1266
}

1267
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1268 1269 1270
{
	static u32 current_clientid = 1;

1271
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1272 1273 1274
	clp->cl_clientid.cl_id = current_clientid++; 
}

1275 1276
static void gen_confirm(struct nfs4_client *clp)
{
1277
	__be32 verf[2];
1278
	static u32 i;
L
Linus Torvalds 已提交
1279

1280 1281 1282
	verf[0] = (__be32)get_seconds();
	verf[1] = (__be32)i++;
	memcpy(clp->cl_confirm.data, verf, sizeof(clp->cl_confirm.data));
L
Linus Torvalds 已提交
1283 1284
}

1285
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1286
{
J
J. Bruce Fields 已提交
1287 1288 1289 1290 1291 1292
	struct nfs4_stid *ret;

	ret = idr_find(&cl->cl_stateids, t->si_opaque.so_id);
	if (!ret || !ret->sc_type)
		return NULL;
	return ret;
J
J. Bruce Fields 已提交
1293 1294
}

1295
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1296 1297
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1298

1299
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1300 1301
	if (!s)
		return NULL;
1302
	if (typemask & s->sc_type)
1303 1304 1305 1306
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1307
static struct nfs4_client *create_client(struct xdr_netobj name,
1308 1309 1310 1311
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1312
	int ret;
1313
	struct net *net = SVC_NET(rqstp);
1314
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1315 1316 1317 1318 1319

	clp = alloc_client(name);
	if (clp == NULL)
		return NULL;

1320
	INIT_LIST_HEAD(&clp->cl_sessions);
1321 1322
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1323
		spin_lock(&nn->client_lock);
1324
		free_client(clp);
1325
		spin_unlock(&nn->client_lock);
1326
		return NULL;
1327
	}
1328
	idr_init(&clp->cl_stateids);
1329
	atomic_set(&clp->cl_refcount, 0);
1330
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1331 1332 1333 1334
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1335
	INIT_LIST_HEAD(&clp->cl_callbacks);
1336
	spin_lock_init(&clp->cl_lock);
1337
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1338
	clp->cl_time = get_seconds();
1339 1340 1341 1342 1343
	clear_bit(0, &clp->cl_cb_slot_busy);
	rpc_init_wait_queue(&clp->cl_cb_waitq, "Backchannel slot table");
	copy_verf(clp, verf);
	rpc_copy_addr((struct sockaddr *) &clp->cl_addr, sa);
	gen_confirm(clp);
1344
	clp->cl_cb_session = NULL;
1345
	clp->net = net;
1346 1347 1348
	return clp;
}

1349
static void
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
add_clp_to_name_tree(struct nfs4_client *new_clp, struct rb_root *root)
{
	struct rb_node **new = &(root->rb_node), *parent = NULL;
	struct nfs4_client *clp;

	while (*new) {
		clp = rb_entry(*new, struct nfs4_client, cl_namenode);
		parent = *new;

		if (compare_blob(&clp->cl_name, &new_clp->cl_name) > 0)
			new = &((*new)->rb_left);
		else
			new = &((*new)->rb_right);
	}

	rb_link_node(&new_clp->cl_namenode, parent, new);
	rb_insert_color(&new_clp->cl_namenode, root);
}

static struct nfs4_client *
find_clp_in_name_tree(struct xdr_netobj *name, struct rb_root *root)
{
	long long cmp;
	struct rb_node *node = root->rb_node;
	struct nfs4_client *clp;

	while (node) {
		clp = rb_entry(node, struct nfs4_client, cl_namenode);
		cmp = compare_blob(&clp->cl_name, name);
		if (cmp > 0)
			node = node->rb_left;
		else if (cmp < 0)
			node = node->rb_right;
		else
			return clp;
	}
	return NULL;
}

static void
add_to_unconfirmed(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1391 1392
{
	unsigned int idhashval;
1393
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1394

1395
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1396
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1397
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1398
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1399
	renew_client(clp);
L
Linus Torvalds 已提交
1400 1401
}

1402
static void
L
Linus Torvalds 已提交
1403 1404 1405
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1406
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1407 1408

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1409
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1410
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1411
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1412
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1413 1414 1415 1416
	renew_client(clp);
}

static struct nfs4_client *
J
J. Bruce Fields 已提交
1417
find_client_in_id_table(struct list_head *tbl, clientid_t *clid, bool sessions)
L
Linus Torvalds 已提交
1418 1419 1420 1421
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

J
J. Bruce Fields 已提交
1422
	list_for_each_entry(clp, &tbl[idhashval], cl_idhash) {
1423
		if (same_clid(&clp->cl_clientid, clid)) {
1424 1425
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1426
			renew_client(clp);
L
Linus Torvalds 已提交
1427
			return clp;
1428
		}
L
Linus Torvalds 已提交
1429 1430 1431 1432
	}
	return NULL;
}

J
J. Bruce Fields 已提交
1433 1434 1435 1436 1437 1438 1439 1440
static struct nfs4_client *
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
{
	struct list_head *tbl = nn->conf_id_hashtbl;

	return find_client_in_id_table(tbl, clid, sessions);
}

L
Linus Torvalds 已提交
1441
static struct nfs4_client *
1442
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1443
{
J
J. Bruce Fields 已提交
1444
	struct list_head *tbl = nn->unconf_id_hashtbl;
L
Linus Torvalds 已提交
1445

J
J. Bruce Fields 已提交
1446
	return find_client_in_id_table(tbl, clid, sessions);
L
Linus Torvalds 已提交
1447 1448
}

1449
static bool clp_used_exchangeid(struct nfs4_client *clp)
1450
{
1451
	return clp->cl_exchange_flags != 0;
1452
} 
1453

1454
static struct nfs4_client *
1455
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1456
{
1457
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1458 1459 1460
}

static struct nfs4_client *
1461
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1462
{
1463
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1464 1465
}

1466
static void
1467
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1468
{
1469
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1470 1471
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
	unsigned short expected_family;

	/* Currently, we only support tcp and tcp6 for the callback channel */
	if (se->se_callback_netid_len == 3 &&
	    !memcmp(se->se_callback_netid_val, "tcp", 3))
		expected_family = AF_INET;
	else if (se->se_callback_netid_len == 4 &&
		 !memcmp(se->se_callback_netid_val, "tcp6", 4))
		expected_family = AF_INET6;
	else
L
Linus Torvalds 已提交
1482 1483
		goto out_err;

1484
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1485
					    se->se_callback_addr_len,
1486 1487
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1488

1489
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1490
		goto out_err;
1491

1492 1493
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1494

1495 1496
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1497
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1498 1499
	return;
out_err:
1500 1501
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1502
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1503 1504 1505 1506 1507 1508
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1509
/*
1510
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1511 1512 1513 1514
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1515 1516
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1517

1518
	dprintk("--> %s slot %p\n", __func__, slot);
1519

1520 1521
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1522

1523
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1524
	if (nfsd4_not_cached(resp)) {
1525
		slot->sl_datalen = 0;
1526
		return;
1527
	}
1528 1529 1530 1531 1532 1533 1534
	slot->sl_datalen = (char *)resp->p - (char *)resp->cstate.datap;
	base = (char *)resp->cstate.datap -
					(char *)resp->xbuf->head[0].iov_base;
	if (read_bytes_from_xdr_buf(resp->xbuf, base, slot->sl_data,
				    slot->sl_datalen))
		WARN("%s: sessions DRC could not cache compound\n", __func__);
	return;
1535 1536 1537
}

/*
1538 1539 1540 1541
 * Encode the replay sequence operation from the slot values.
 * If cachethis is FALSE encode the uncached rep error on the next
 * operation which sets resp->p and increments resp->opcnt for
 * nfs4svc_encode_compoundres.
1542 1543
 *
 */
1544 1545 1546
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1547
{
1548 1549
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1550

1551 1552 1553
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1554

1555
	/* Return nfserr_retry_uncached_rep in next operation. */
1556
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1557 1558 1559
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1560
	}
1561
	return op->status;
1562 1563 1564
}

/*
1565 1566
 * The sequence operation is not cached because we can use the slot and
 * session values.
1567 1568
 */
__be32
1569 1570
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1571
{
1572
	struct nfsd4_slot *slot = resp->cstate.slot;
1573 1574
	__be32 status;

1575
	dprintk("--> %s slot %p\n", __func__, slot);
1576

1577 1578 1579 1580
	/* Either returns 0 or nfserr_retry_uncached */
	status = nfsd4_enc_sequence_replay(resp->rqstp->rq_argp, resp);
	if (status == nfserr_retry_uncached_rep)
		return status;
1581

1582 1583
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1584

1585 1586 1587
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1588 1589 1590 1591

	return status;
}

A
Andy Adamson 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
/*
 * Set the exchange_id flags returned by the server.
 */
static void
nfsd4_set_ex_flags(struct nfs4_client *new, struct nfsd4_exchange_id *clid)
{
	/* pNFS is not supported */
	new->cl_exchange_flags |= EXCHGID4_FLAG_USE_NON_PNFS;

	/* Referrals are supported, Migration is not. */
	new->cl_exchange_flags |= EXCHGID4_FLAG_SUPP_MOVED_REFER;

	/* set the wire flags to return to client. */
	clid->flags = new->cl_exchange_flags;
}

1608 1609 1610 1611 1612 1613 1614 1615
static bool client_has_state(struct nfs4_client *clp)
{
	/*
	 * Note clp->cl_openowners check isn't quite right: there's no
	 * need to count owners without stateid's.
	 *
	 * Also note we should probably be using this in 4.0 case too.
	 */
1616 1617 1618
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1619 1620
}

A
Andy Adamson 已提交
1621 1622 1623 1624 1625
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1626
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1627
	__be32 status;
1628
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1629
	nfs4_verifier		verf = exid->verifier;
1630
	struct sockaddr		*sa = svc_addr(rqstp);
1631
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1632
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1633

1634
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1635
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1636
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1637
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1638
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1639

1640
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1641 1642 1643 1644 1645 1646
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1647 1648
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1649 1650 1651 1652 1653
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1654
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1655
	nfs4_lock_state();
1656
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1657
	if (conf) {
1658 1659 1660
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1661 1662
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1663
				status = nfserr_inval;
1664 1665
				goto out;
			}
1666
			if (!creds_match) { /* case 9 */
1667
				status = nfserr_perm;
1668 1669 1670
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1671 1672 1673
				status = nfserr_not_same;
				goto out;
			}
1674 1675 1676 1677
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1678
		}
1679
		if (!creds_match) { /* case 3 */
1680 1681
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1682 1683 1684 1685 1686
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1687
		if (verfs_match) { /* case 2 */
1688
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1689 1690 1691 1692 1693
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1694 1695
	}

1696
	if (update) { /* case 7 */
1697 1698
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1699 1700
	}

1701
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1702
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1703 1704
		expire_client(unconf);

1705
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1706
out_new:
J
Jeff Layton 已提交
1707
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1708
	if (new == NULL) {
1709
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1710 1711
		goto out;
	}
1712
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1713

1714
	gen_clid(new, nn);
1715
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1716 1717 1718 1719
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1720
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1721 1722 1723
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1724
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1725 1726 1727 1728 1729
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1730 1731
}

J
J. Bruce Fields 已提交
1732
static __be32
1733
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1734
{
1735 1736
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1737 1738

	/* The slot is in use, and no response has been sent. */
1739 1740
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1741 1742 1743 1744
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1745
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1746
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1747
		return nfs_ok;
1748
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1749 1750 1751 1752
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1753 1754 1755 1756 1757 1758 1759
/*
 * Cache the create session result into the create session single DRC
 * slot cache by saving the xdr structure. sl_seqid has been set.
 * Do this for solo or embedded create session operations.
 */
static void
nfsd4_cache_create_session(struct nfsd4_create_session *cr_ses,
J
J. Bruce Fields 已提交
1760
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
{
	slot->sl_status = nfserr;
	memcpy(&slot->sl_cr_ses, cr_ses, sizeof(*cr_ses));
}

static __be32
nfsd4_replay_create_session(struct nfsd4_create_session *cr_ses,
			    struct nfsd4_clid_slot *slot)
{
	memcpy(cr_ses, &slot->sl_cr_ses, sizeof(*cr_ses));
	return slot->sl_status;
}

1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
#define NFSD_MIN_REQ_HDR_SEQ_SZ	((\
			2 * 2 + /* credential,verifier: AUTH_NULL, length 0 */ \
			1 +	/* MIN tag is length with zero, only length */ \
			3 +	/* version, opcount, opcode */ \
			XDR_QUADLEN(NFS4_MAX_SESSIONID_LEN) + \
				/* seqid, slotID, slotID, cache */ \
			4 ) * sizeof(__be32))

#define NFSD_MIN_RESP_HDR_SEQ_SZ ((\
			2 +	/* verifier: AUTH_NULL, length 0 */\
			1 +	/* status */ \
			1 +	/* MIN tag is length with zero, only length */ \
			3 +	/* opcount, opcode, opstatus*/ \
			XDR_QUADLEN(NFS4_MAX_SESSIONID_LEN) + \
				/* seqid, slotID, slotID, slotID, status */ \
			5 ) * sizeof(__be32))

J
J. Bruce Fields 已提交
1791
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1792 1793 1794 1795 1796
{
	return fchannel.maxreq_sz < NFSD_MIN_REQ_HDR_SEQ_SZ
		|| fchannel.maxresp_sz < NFSD_MIN_RESP_HDR_SEQ_SZ;
}

A
Andy Adamson 已提交
1797 1798 1799 1800 1801
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1802
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1803
	struct nfs4_client *conf, *unconf;
1804
	struct nfsd4_session *new;
1805
	struct nfsd4_conn *conn;
1806
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1807
	__be32 status = 0;
1808
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1809

1810 1811
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1812 1813
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1814
	new = alloc_session(&cr_ses->fore_channel, nn);
1815 1816 1817 1818 1819 1820
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1821

A
Andy Adamson 已提交
1822
	nfs4_lock_state();
1823
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1824
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
1825
	WARN_ON_ONCE(conf && unconf);
A
Andy Adamson 已提交
1826 1827

	if (conf) {
1828 1829
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1830
		if (status == nfserr_replay_cache) {
1831
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1832
			goto out_free_conn;
1833
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1834
			status = nfserr_seq_misordered;
1835
			goto out_free_conn;
A
Andy Adamson 已提交
1836 1837
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1838
		struct nfs4_client *old;
A
Andy Adamson 已提交
1839
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1840
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1841
			status = nfserr_clid_inuse;
1842
			goto out_free_conn;
A
Andy Adamson 已提交
1843
		}
1844 1845
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1846 1847
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1848
			status = nfserr_seq_misordered;
1849
			goto out_free_conn;
A
Andy Adamson 已提交
1850
		}
1851
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
1852 1853 1854 1855
		if (old) {
			status = mark_client_expired(old);
			if (status)
				goto out_free_conn;
J
J. Bruce Fields 已提交
1856
			expire_client(old);
1857
		}
J
J. Bruce Fields 已提交
1858
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1859 1860 1861
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1862
		goto out_free_conn;
A
Andy Adamson 已提交
1863
	}
1864
	status = nfs_ok;
1865 1866 1867 1868 1869 1870
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1871 1872 1873
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1874
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1875
	       NFS4_MAX_SESSIONID_LEN);
1876 1877
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1878
	cs_slot->sl_seqid++;
1879
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1880

1881 1882
	/* cache solo and embedded create sessions under the state lock */
	nfsd4_cache_create_session(cr_ses, cs_slot, status);
A
Andy Adamson 已提交
1883 1884
	nfs4_unlock_state();
	return status;
1885
out_free_conn:
1886
	nfs4_unlock_state();
1887 1888 1889
	free_conn(conn);
out_free_session:
	__free_session(new);
J
J. Bruce Fields 已提交
1890
	return status;
A
Andy Adamson 已提交
1891 1892
}

1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906
static __be32 nfsd4_map_bcts_dir(u32 *dir)
{
	switch (*dir) {
	case NFS4_CDFC4_FORE:
	case NFS4_CDFC4_BACK:
		return nfs_ok;
	case NFS4_CDFC4_FORE_OR_BOTH:
	case NFS4_CDFC4_BACK_OR_BOTH:
		*dir = NFS4_CDFC4_BOTH;
		return nfs_ok;
	};
	return nfserr_inval;
}

1907 1908 1909
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1910
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1911

1912
	spin_lock(&nn->client_lock);
1913 1914
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1915
	spin_unlock(&nn->client_lock);
1916 1917 1918 1919 1920 1921

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1922 1923 1924 1925 1926
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1927
	struct nfsd4_conn *conn;
1928
	struct nfsd4_session *session;
1929
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1930 1931 1932

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1933
	nfs4_lock_state();
1934
	spin_lock(&nn->client_lock);
1935
	session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1936
	spin_unlock(&nn->client_lock);
1937 1938 1939
	status = nfserr_badsession;
	if (!session)
		goto out;
1940
	status = nfsd4_map_bcts_dir(&bcts->dir);
1941
	if (status)
1942
		goto out;
1943
	conn = alloc_conn(rqstp, bcts->dir);
1944
	status = nfserr_jukebox;
1945
	if (!conn)
1946 1947 1948 1949 1950 1951
		goto out;
	nfsd4_init_conn(rqstp, conn, session);
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
1952 1953
}

1954 1955 1956 1957 1958 1959 1960
static bool nfsd4_compound_in_session(struct nfsd4_session *session, struct nfs4_sessionid *sid)
{
	if (!session)
		return 0;
	return !memcmp(sid, &session->se_sessionid, sizeof(*sid));
}

A
Andy Adamson 已提交
1961 1962 1963 1964 1965
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1966
	struct nfsd4_session *ses;
1967
	__be32 status;
1968
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1969

1970 1971
	nfs4_lock_state();
	status = nfserr_not_only_op;
1972
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1973
		if (!nfsd4_last_compound_op(r))
1974
			goto out;
1975
	}
B
Benny Halevy 已提交
1976
	dump_sessionid(__func__, &sessionid->sessionid);
1977
	spin_lock(&nn->client_lock);
1978
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
1979 1980 1981
	status = nfserr_badsession;
	if (!ses)
		goto out_client_lock;
1982 1983 1984
	status = mark_session_dead_locked(ses);
	if (status)
		goto out_client_lock;
B
Benny Halevy 已提交
1985
	unhash_session(ses);
1986
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1987

1988
	nfsd4_probe_callback_sync(ses->se_client);
1989

1990
	spin_lock(&nn->client_lock);
1991
	free_session(ses);
B
Benny Halevy 已提交
1992
	status = nfs_ok;
1993 1994
out_client_lock:
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1995
out:
1996
	nfs4_unlock_state();
B
Benny Halevy 已提交
1997
	return status;
A
Andy Adamson 已提交
1998 1999
}

2000
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
2001 2002 2003 2004
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
2005
		if (c->cn_xprt == xpt) {
2006 2007 2008 2009 2010 2011
			return c;
		}
	}
	return NULL;
}

2012
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
2013 2014
{
	struct nfs4_client *clp = ses->se_client;
2015
	struct nfsd4_conn *c;
2016
	int ret;
2017 2018

	spin_lock(&clp->cl_lock);
2019
	c = __nfsd4_find_conn(new->cn_xprt, ses);
2020 2021 2022 2023 2024 2025 2026
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
2027 2028 2029 2030
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2031 2032 2033
	return;
}

2034 2035 2036 2037 2038 2039 2040
static bool nfsd4_session_too_many_ops(struct svc_rqst *rqstp, struct nfsd4_session *session)
{
	struct nfsd4_compoundargs *args = rqstp->rq_argp;

	return args->opcnt > session->se_fchannel.maxops;
}

M
Mi Jinlong 已提交
2041 2042 2043 2044 2045 2046 2047 2048
static bool nfsd4_request_too_big(struct svc_rqst *rqstp,
				  struct nfsd4_session *session)
{
	struct xdr_buf *xb = &rqstp->rq_arg;

	return xb->len > session->se_fchannel.maxreq_sz;
}

A
Andy Adamson 已提交
2049
__be32
B
Benny Halevy 已提交
2050
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2051 2052 2053
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2054
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2055
	struct nfsd4_session *session;
2056
	struct nfs4_client *clp;
B
Benny Halevy 已提交
2057
	struct nfsd4_slot *slot;
2058
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2059
	__be32 status;
2060
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2061

2062 2063 2064
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2065 2066 2067 2068 2069 2070 2071 2072
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2073
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2074
	status = nfserr_badsession;
2075
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2076
	if (!session)
2077 2078 2079 2080 2081
		goto out_no_session;
	clp = session->se_client;
	status = get_client_locked(clp);
	if (status)
		goto out_no_session;
2082 2083 2084
	status = nfsd4_get_session_locked(session);
	if (status)
		goto out_put_client;
B
Benny Halevy 已提交
2085

2086 2087
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
2088
		goto out_put_session;
2089

M
Mi Jinlong 已提交
2090 2091
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
2092
		goto out_put_session;
M
Mi Jinlong 已提交
2093

B
Benny Halevy 已提交
2094
	status = nfserr_badslot;
2095
	if (seq->slotid >= session->se_fchannel.maxreqs)
2096
		goto out_put_session;
B
Benny Halevy 已提交
2097

2098
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2099 2100
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2101 2102 2103 2104 2105
	/* We do not negotiate the number of slots yet, so set the
	 * maxslots to the session maxreqs which is used to encode
	 * sr_highest_slotid and the sr_target_slot id to maxslots */
	seq->maxslots = session->se_fchannel.maxreqs;

2106 2107
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2108
	if (status == nfserr_replay_cache) {
2109 2110
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
2111
			goto out_put_session;
B
Benny Halevy 已提交
2112 2113
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2114
		/* Return the cached reply status and set cstate->status
2115
		 * for nfsd4_proc_compound processing */
2116
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2117
		cstate->status = nfserr_replay_cache;
2118
		goto out;
B
Benny Halevy 已提交
2119 2120
	}
	if (status)
2121
		goto out_put_session;
B
Benny Halevy 已提交
2122

2123 2124
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2125

B
Benny Halevy 已提交
2126 2127
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2128
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2129 2130
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2131 2132
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2133 2134 2135 2136 2137

	cstate->slot = slot;
	cstate->session = session;

out:
2138 2139 2140 2141 2142 2143 2144 2145 2146
	switch (clp->cl_cb_state) {
	case NFSD4_CB_DOWN:
		seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
		break;
	case NFSD4_CB_FAULT:
		seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
		break;
	default:
		seq->status_flags = 0;
2147
	}
2148
out_no_session:
2149
	kfree(conn);
2150
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2151
	return status;
2152 2153
out_put_session:
	nfsd4_put_session(session);
2154 2155 2156
out_put_client:
	put_client_renew_locked(clp);
	goto out_no_session;
A
Andy Adamson 已提交
2157 2158
}

2159 2160 2161 2162
__be32
nfsd4_destroy_clientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_destroy_clientid *dc)
{
	struct nfs4_client *conf, *unconf, *clp;
J
J. Bruce Fields 已提交
2163
	__be32 status = 0;
2164
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2165 2166

	nfs4_lock_state();
2167
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2168
	conf = find_confirmed_client(&dc->clientid, true, nn);
2169
	WARN_ON_ONCE(conf && unconf);
2170 2171 2172 2173

	if (conf) {
		clp = conf;

2174
		if (client_has_state(conf)) {
2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190
			status = nfserr_clientid_busy;
			goto out;
		}
	} else if (unconf)
		clp = unconf;
	else {
		status = nfserr_stale_clientid;
		goto out;
	}

	expire_client(clp);
out:
	nfs4_unlock_state();
	return status;
}

2191 2192 2193
__be32
nfsd4_reclaim_complete(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_reclaim_complete *rc)
{
J
J. Bruce Fields 已提交
2194
	__be32 status = 0;
2195

2196 2197 2198 2199 2200 2201 2202 2203 2204
	if (rc->rca_one_fs) {
		if (!cstate->current_fh.fh_dentry)
			return nfserr_nofilehandle;
		/*
		 * We don't take advantage of the rca_one_fs case.
		 * That's OK, it's optional, we can safely ignore it.
		 */
		 return nfs_ok;
	}
2205

2206
	nfs4_lock_state();
2207
	status = nfserr_complete_already;
2208 2209
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2210 2211 2212 2213
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2214 2215 2216 2217 2218 2219 2220
		/*
		 * The following error isn't really legal.
		 * But we only get here if the client just explicitly
		 * destroyed the client.  Surely it no longer cares what
		 * error it gets back on an operation for the dead
		 * client.
		 */
2221 2222 2223
		goto out;

	status = nfs_ok;
2224
	nfsd4_client_record_create(cstate->session->se_client);
2225
out:
2226
	nfs4_unlock_state();
2227
	return status;
2228 2229
}

2230
__be32
2231 2232
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2233
{
2234
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2235
	nfs4_verifier		clverifier = setclid->se_verf;
2236
	struct nfs4_client	*conf, *unconf, *new;
2237
	__be32 			status;
2238 2239
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2240
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2241
	nfs4_lock_state();
2242
	conf = find_confirmed_client_by_name(&clname, nn);
2243
	if (conf) {
2244
		/* case 0: */
L
Linus Torvalds 已提交
2245
		status = nfserr_clid_inuse;
2246 2247
		if (clp_used_exchangeid(conf))
			goto out;
2248
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2249 2250 2251 2252 2253
			char addr_str[INET6_ADDRSTRLEN];
			rpc_ntop((struct sockaddr *) &conf->cl_addr, addr_str,
				 sizeof(addr_str));
			dprintk("NFSD: setclientid: string in use by client "
				"at %s\n", addr_str);
L
Linus Torvalds 已提交
2254 2255 2256
			goto out;
		}
	}
2257
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2258 2259
	if (unconf)
		expire_client(unconf);
2260
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2261
	new = create_client(clname, rqstp, &clverifier);
2262 2263
	if (new == NULL)
		goto out;
2264
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2265
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2266
		copy_clid(new, conf);
2267
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2268
		gen_clid(new, nn);
2269
	new->cl_minorversion = 0;
2270
	gen_callback(new, setclid, rqstp);
2271
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2272 2273 2274 2275 2276 2277 2278 2279 2280 2281
	setclid->se_clientid.cl_boot = new->cl_clientid.cl_boot;
	setclid->se_clientid.cl_id = new->cl_clientid.cl_id;
	memcpy(setclid->se_confirm.data, new->cl_confirm.data, sizeof(setclid->se_confirm.data));
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}


2282
__be32
2283 2284 2285
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2286
{
2287
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2288 2289
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2290
	__be32 status;
2291
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2292

2293
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2294 2295
		return nfserr_stale_clientid;
	nfs4_lock_state();
2296

2297
	conf = find_confirmed_client(clid, false, nn);
2298
	unconf = find_unconfirmed_client(clid, false, nn);
2299
	/*
2300 2301 2302 2303
	 * We try hard to give out unique clientid's, so if we get an
	 * attempt to confirm the same clientid with a different cred,
	 * there's a bug somewhere.  Let's charitably assume it's our
	 * bug.
2304
	 */
2305 2306 2307 2308 2309
	status = nfserr_serverfault;
	if (unconf && !same_creds(&unconf->cl_cred, &rqstp->rq_cred))
		goto out;
	if (conf && !same_creds(&conf->cl_cred, &rqstp->rq_cred))
		goto out;
2310
	/* cases below refer to rfc 3530 section 14.2.34: */
2311 2312
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2313
			status = nfs_ok;
2314 2315 2316 2317 2318 2319
		else /* case 4: client hasn't noticed we rebooted yet? */
			status = nfserr_stale_clientid;
		goto out;
	}
	status = nfs_ok;
	if (conf) { /* case 1: callback update */
2320 2321 2322
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2323
	} else { /* case 3: normal case; new or rebooted client */
2324
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2325 2326 2327 2328
		if (conf) {
			status = mark_client_expired(conf);
			if (status)
				goto out;
2329
			expire_client(conf);
2330
		}
2331
		move_to_confirmed(unconf);
2332
		nfsd4_probe_callback(unconf);
2333
	}
L
Linus Torvalds 已提交
2334 2335 2336 2337 2338
out:
	nfs4_unlock_state();
	return status;
}

2339 2340 2341 2342 2343
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2344
/* OPEN Share state helper functions */
2345
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2346 2347 2348
{
	unsigned int hashval = file_hashval(ino);

2349 2350 2351 2352 2353 2354 2355 2356 2357
	atomic_set(&fp->fi_ref, 1);
	INIT_LIST_HEAD(&fp->fi_stateids);
	INIT_LIST_HEAD(&fp->fi_delegations);
	fp->fi_inode = igrab(ino);
	fp->fi_had_conflict = false;
	fp->fi_lease = NULL;
	memset(fp->fi_fds, 0, sizeof(fp->fi_fds));
	memset(fp->fi_access, 0, sizeof(fp->fi_access));
	spin_lock(&recall_lock);
2358
	hlist_add_head(&fp->fi_hash, &file_hashtbl[hashval]);
2359
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2360 2361
}

N
NeilBrown 已提交
2362
static void
2363
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2364
{
N
NeilBrown 已提交
2365 2366
	if (*slab == NULL)
		return;
2367
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2368
	*slab = NULL;
L
Linus Torvalds 已提交
2369 2370
}

2371
void
L
Linus Torvalds 已提交
2372 2373
nfsd4_free_slabs(void)
{
2374 2375
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2376
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2377
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2378
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2379
}
L
Linus Torvalds 已提交
2380

2381
int
N
NeilBrown 已提交
2382 2383
nfsd4_init_slabs(void)
{
2384 2385 2386 2387 2388
	openowner_slab = kmem_cache_create("nfsd4_openowners",
			sizeof(struct nfs4_openowner), 0, 0, NULL);
	if (openowner_slab == NULL)
		goto out_nomem;
	lockowner_slab = kmem_cache_create("nfsd4_lockowners",
2389
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2390
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2391 2392
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2393
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2394 2395
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2396
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2397
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2398 2399
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2400
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2401
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2402 2403
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2404 2405 2406 2407 2408
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2409 2410
}

2411 2412 2413 2414 2415 2416 2417
void nfs4_free_openowner(struct nfs4_openowner *oo)
{
	kfree(oo->oo_owner.so_owner.data);
	kmem_cache_free(openowner_slab, oo);
}

void nfs4_free_lockowner(struct nfs4_lockowner *lo)
L
Linus Torvalds 已提交
2418
{
2419 2420
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2421 2422
}

2423
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2424
{
2425 2426 2427
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2428 2429
}

2430
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2431
{
L
Linus Torvalds 已提交
2432 2433
	struct nfs4_stateowner *sop;

2434
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2435 2436 2437 2438 2439
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2440
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2441
		return NULL;
2442 2443 2444
	}
	sop->so_owner.len = owner->len;

2445
	INIT_LIST_HEAD(&sop->so_stateids);
2446 2447 2448 2449 2450
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2451
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2452
{
2453 2454 2455
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2456
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2457 2458
}

2459
static struct nfs4_openowner *
2460
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2461
	struct nfs4_openowner *oo;
2462

2463 2464
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2465
		return NULL;
2466 2467
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2468
	oo->oo_flags = NFS4_OO_NEW;
2469
	oo->oo_time = 0;
2470
	oo->oo_last_closed_stid = NULL;
2471 2472 2473
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2474 2475
}

2476
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2477
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2478

J
J. Bruce Fields 已提交
2479
	stp->st_stid.sc_type = NFS4_OPEN_STID;
2480
	INIT_LIST_HEAD(&stp->st_lockowners);
2481
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2482
	list_add(&stp->st_perfile, &fp->fi_stateids);
2483
	stp->st_stateowner = &oo->oo_owner;
2484
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2485 2486 2487
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2488
	set_access(open->op_share_access, stp);
2489
	set_deny(open->op_share_deny, stp);
2490
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2491 2492
}

2493
static void
2494
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2495
{
2496 2497
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2498
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2499

2500
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2501
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2502 2503 2504
}

static int
2505 2506 2507 2508 2509 2510
same_owner_str(struct nfs4_stateowner *sop, struct xdr_netobj *owner,
							clientid_t *clid)
{
	return (sop->so_owner.len == owner->len) &&
		0 == memcmp(sop->so_owner.data, owner->data, owner->len) &&
		(sop->so_client->cl_clientid.cl_id == clid->cl_id);
L
Linus Torvalds 已提交
2511 2512
}

2513
static struct nfs4_openowner *
2514 2515
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2516
{
2517 2518
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2519
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2520

2521
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2522 2523
		if (!so->so_is_open_owner)
			continue;
2524 2525
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2526 2527 2528
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2529 2530 2531
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
	}
	return NULL;
}

/* search file_hashtbl[] for file */
static struct nfs4_file *
find_file(struct inode *ino)
{
	unsigned int hashval = file_hashval(ino);
	struct nfs4_file *fp;

2543
	spin_lock(&recall_lock);
2544
	hlist_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2545 2546
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2547
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2548
			return fp;
2549
		}
L
Linus Torvalds 已提交
2550
	}
2551
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2552 2553 2554 2555 2556 2557 2558
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2559
static __be32
L
Linus Torvalds 已提交
2560 2561 2562 2563
nfs4_share_conflict(struct svc_fh *current_fh, unsigned int deny_type)
{
	struct inode *ino = current_fh->fh_dentry->d_inode;
	struct nfs4_file *fp;
2564
	struct nfs4_ol_stateid *stp;
2565
	__be32 ret;
L
Linus Torvalds 已提交
2566 2567

	fp = find_file(ino);
2568 2569
	if (!fp)
		return nfs_ok;
2570
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2571
	/* Search for conflicting share reservations */
2572
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2573 2574
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2575
			goto out;
L
Linus Torvalds 已提交
2576
	}
2577 2578 2579 2580
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2581 2582
}

2583
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2584 2585 2586 2587 2588 2589 2590 2591 2592 2593
{
	/* We're assuming the state code never drops its reference
	 * without first removing the lease.  Since we're in this lease
	 * callback (and since the lease code is serialized by the kernel
	 * lock) we know the server hasn't removed the lease yet, we know
	 * it's safe to take a reference: */
	atomic_inc(&dp->dl_count);

	list_add_tail(&dp->dl_recall_lru, &del_recall_lru);

2594
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2595 2596
	dp->dl_time = get_seconds();

2597 2598 2599
	nfsd4_cb_recall(dp);
}

2600
/* Called from break_lease() with lock_flocks() held. */
2601 2602
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2603 2604
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2605

2606 2607 2608 2609 2610 2611 2612 2613
	if (!fp) {
		WARN(1, "(%p)->fl_owner NULL\n", fl);
		return;
	}
	if (fp->fi_had_conflict) {
		WARN(1, "duplicate break on %p\n", fp);
		return;
	}
2614 2615
	/*
	 * We don't want the locks code to timeout the lease for us;
2616
	 * we'll remove it ourself if a delegation isn't returned
2617
	 * in time:
2618 2619
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2620

2621
	spin_lock(&recall_lock);
2622 2623 2624
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2625
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636
}

static
int nfsd_change_deleg_cb(struct file_lock **onlist, int arg)
{
	if (arg & F_UNLCK)
		return lease_modify(onlist, arg);
	else
		return -EAGAIN;
}

2637
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2638 2639
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2640 2641
};

2642 2643 2644 2645 2646 2647 2648 2649 2650 2651
static __be32 nfsd4_check_seqid(struct nfsd4_compound_state *cstate, struct nfs4_stateowner *so, u32 seqid)
{
	if (nfsd4_has_session(cstate))
		return nfs_ok;
	if (seqid == so->so_seqid - 1)
		return nfserr_replay_me;
	if (seqid == so->so_seqid)
		return nfs_ok;
	return nfserr_bad_seqid;
}
L
Linus Torvalds 已提交
2652

2653
__be32
A
Andy Adamson 已提交
2654
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2655
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2656 2657 2658 2659
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2660
	struct nfs4_openowner *oo = NULL;
2661
	__be32 status;
L
Linus Torvalds 已提交
2662

2663
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2664
		return nfserr_stale_clientid;
2665 2666 2667 2668 2669 2670 2671
	/*
	 * In case we need it later, after we've already created the
	 * file and don't want to risk a further failure:
	 */
	open->op_file = nfsd4_alloc_file();
	if (open->op_file == NULL)
		return nfserr_jukebox;
L
Linus Torvalds 已提交
2672

2673
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2674
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2675 2676
	open->op_openowner = oo;
	if (!oo) {
2677 2678
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2679
		if (clp == NULL)
2680
			return nfserr_expired;
2681
		goto new_owner;
L
Linus Torvalds 已提交
2682
	}
2683
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2684
		/* Replace unconfirmed owners without checking for replay. */
2685 2686 2687
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2688
		goto new_owner;
2689
	}
2690 2691 2692 2693 2694
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2695 2696 2697 2698 2699
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2700 2701 2702 2703
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2704
	return nfs_ok;
L
Linus Torvalds 已提交
2705 2706
}

2707
static inline __be32
N
NeilBrown 已提交
2708 2709 2710 2711 2712 2713 2714 2715
nfs4_check_delegmode(struct nfs4_delegation *dp, int flags)
{
	if ((flags & WR_STATE) && (dp->dl_type == NFS4_OPEN_DELEGATE_READ))
		return nfserr_openmode;
	else
		return nfs_ok;
}

2716
static int share_access_to_flags(u32 share_access)
2717
{
2718
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2719 2720
}

2721
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2722
{
2723
	struct nfs4_stid *ret;
2724

2725
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2726 2727 2728
	if (!ret)
		return NULL;
	return delegstateid(ret);
2729 2730
}

2731 2732 2733 2734 2735 2736
static bool nfsd4_is_deleg_cur(struct nfsd4_open *open)
{
	return open->op_claim_type == NFS4_OPEN_CLAIM_DELEGATE_CUR ||
	       open->op_claim_type == NFS4_OPEN_CLAIM_DELEG_CUR_FH;
}

2737
static __be32
2738
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2739 2740 2741
		struct nfs4_delegation **dp)
{
	int flags;
2742
	__be32 status = nfserr_bad_stateid;
2743

2744
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2745
	if (*dp == NULL)
2746
		goto out;
2747
	flags = share_access_to_flags(open->op_share_access);
2748 2749 2750
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2751
out:
2752
	if (!nfsd4_is_deleg_cur(open))
2753 2754 2755
		return nfs_ok;
	if (status)
		return status;
2756
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2757
	return nfs_ok;
2758 2759
}

2760
static __be32
2761
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2762
{
2763
	struct nfs4_ol_stateid *local;
2764
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2765

2766
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2767 2768 2769 2770
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2771
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2772 2773 2774
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2775
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2776
	}
2777
	return nfs_ok;
L
Linus Torvalds 已提交
2778 2779
}

2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
static inline int nfs4_access_to_access(u32 nfs4_access)
{
	int flags = 0;

	if (nfs4_access & NFS4_SHARE_ACCESS_READ)
		flags |= NFSD_MAY_READ;
	if (nfs4_access & NFS4_SHARE_ACCESS_WRITE)
		flags |= NFSD_MAY_WRITE;
	return flags;
}

2791 2792
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2793 2794
{
	__be32 status;
2795 2796 2797
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808
	if (!fp->fi_fds[oflag]) {
		status = nfsd_open(rqstp, cur_fh, S_IFREG, access,
			&fp->fi_fds[oflag]);
		if (status)
			return status;
	}
	nfs4_file_get_access(fp, oflag);

	return nfs_ok;
}

2809
static inline __be32
L
Linus Torvalds 已提交
2810 2811 2812 2813 2814 2815 2816 2817 2818 2819
nfsd4_truncate(struct svc_rqst *rqstp, struct svc_fh *fh,
		struct nfsd4_open *open)
{
	struct iattr iattr = {
		.ia_valid = ATTR_SIZE,
		.ia_size = 0,
	};
	if (!open->op_truncate)
		return 0;
	if (!(open->op_share_access & NFS4_SHARE_ACCESS_WRITE))
2820
		return nfserr_inval;
L
Linus Torvalds 已提交
2821 2822 2823
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2824
static __be32
2825
nfs4_upgrade_open(struct svc_rqst *rqstp, struct nfs4_file *fp, struct svc_fh *cur_fh, struct nfs4_ol_stateid *stp, struct nfsd4_open *open)
L
Linus Torvalds 已提交
2826
{
2827
	u32 op_share_access = open->op_share_access;
2828
	bool new_access;
2829
	__be32 status;
L
Linus Torvalds 已提交
2830

2831
	new_access = !test_access(op_share_access, stp);
2832
	if (new_access) {
2833
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2834 2835
		if (status)
			return status;
2836
	}
L
Linus Torvalds 已提交
2837 2838
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2839
		if (new_access) {
2840
			int oflag = nfs4_access_to_omode(op_share_access);
2841 2842
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2843 2844 2845
		return status;
	}
	/* remember the open */
2846
	set_access(op_share_access, stp);
2847
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2848 2849 2850 2851 2852 2853

	return nfs_ok;
}


static void
2854
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2855
{
2856
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2857 2858
}

2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871
/* Should we give out recallable state?: */
static bool nfsd4_cb_channel_good(struct nfs4_client *clp)
{
	if (clp->cl_cb_state == NFSD4_CB_UP)
		return true;
	/*
	 * In the sessions case, since we don't have to establish a
	 * separate connection for callbacks, we assume it's OK
	 * until we hear otherwise:
	 */
	return clp->cl_minorversion && clp->cl_cb_state == NFSD4_CB_UNKNOWN;
}

2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
static struct file_lock *nfs4_alloc_init_lease(struct nfs4_delegation *dp, int flag)
{
	struct file_lock *fl;

	fl = locks_alloc_lock();
	if (!fl)
		return NULL;
	locks_init_lock(fl);
	fl->fl_lmops = &nfsd_lease_mng_ops;
	fl->fl_flags = FL_LEASE;
	fl->fl_type = flag == NFS4_OPEN_DELEGATE_READ? F_RDLCK: F_WRLCK;
	fl->fl_end = OFFSET_MAX;
2884
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2885 2886 2887 2888
	fl->fl_pid = current->tgid;
	return fl;
}

2889 2890
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2891
	struct nfs4_file *fp = dp->dl_file;
2892 2893 2894 2895 2896 2897
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2898
	fl->fl_file = find_readable_file(fp);
2899
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2900
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2901
	if (status) {
2902
		list_del_init(&dp->dl_perclnt);
2903 2904 2905
		locks_free_lock(fl);
		return -ENOMEM;
	}
2906
	fp->fi_lease = fl;
A
Al Viro 已提交
2907
	fp->fi_deleg_file = get_file(fl->fl_file);
2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
	atomic_set(&fp->fi_delegees, 1);
	list_add(&dp->dl_perfile, &fp->fi_delegations);
	return 0;
}

static int nfs4_set_delegation(struct nfs4_delegation *dp, int flag)
{
	struct nfs4_file *fp = dp->dl_file;

	if (!fp->fi_lease)
		return nfs4_setlease(dp, flag);
	spin_lock(&recall_lock);
	if (fp->fi_had_conflict) {
		spin_unlock(&recall_lock);
		return -EAGAIN;
	}
	atomic_inc(&fp->fi_delegees);
	list_add(&dp->dl_perfile, &fp->fi_delegations);
	spin_unlock(&recall_lock);
2927
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2928 2929 2930
	return 0;
}

2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946
static void nfsd4_open_deleg_none_ext(struct nfsd4_open *open, int status)
{
	open->op_delegate_type = NFS4_OPEN_DELEGATE_NONE_EXT;
	if (status == -EAGAIN)
		open->op_why_no_deleg = WND4_CONTENTION;
	else {
		open->op_why_no_deleg = WND4_RESOURCE;
		switch (open->op_deleg_want) {
		case NFS4_SHARE_WANT_READ_DELEG:
		case NFS4_SHARE_WANT_WRITE_DELEG:
		case NFS4_SHARE_WANT_ANY_DELEG:
			break;
		case NFS4_SHARE_WANT_CANCEL:
			open->op_why_no_deleg = WND4_CANCELLED;
			break;
		case NFS4_SHARE_WANT_NO_DELEG:
2947
			WARN_ON_ONCE(1);
2948 2949 2950 2951
		}
	}
}

L
Linus Torvalds 已提交
2952 2953 2954 2955
/*
 * Attempt to hand out a delegation.
 */
static void
2956 2957
nfs4_open_delegation(struct net *net, struct svc_fh *fh,
		     struct nfsd4_open *open, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
2958 2959
{
	struct nfs4_delegation *dp;
2960
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2961
	int cb_up;
2962
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2963

2964
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2965
	flag = NFS4_OPEN_DELEGATE_NONE;
2966 2967 2968
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2969
			if (!cb_up)
2970 2971 2972 2973 2974 2975 2976 2977
				open->op_recall = 1;
			flag = open->op_delegate_type;
			if (flag == NFS4_OPEN_DELEGATE_NONE)
				goto out;
			break;
		case NFS4_OPEN_CLAIM_NULL:
			/* Let's not give out any delegations till everyone's
			 * had the chance to reclaim theirs.... */
2978
			if (locks_in_grace(net))
2979
				goto out;
2980
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2981 2982 2983 2984 2985 2986 2987 2988 2989
				goto out;
			if (open->op_share_access & NFS4_SHARE_ACCESS_WRITE)
				flag = NFS4_OPEN_DELEGATE_WRITE;
			else
				flag = NFS4_OPEN_DELEGATE_READ;
			break;
		default:
			goto out;
	}
L
Linus Torvalds 已提交
2990

2991
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2992 2993
	if (dp == NULL)
		goto out_no_deleg;
2994
	status = nfs4_set_delegation(dp, flag);
2995
	if (status)
2996
		goto out_free;
L
Linus Torvalds 已提交
2997

2998
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2999

3000
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
3001
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
3002 3003
out:
	open->op_delegate_type = flag;
3004 3005 3006 3007 3008
	if (flag == NFS4_OPEN_DELEGATE_NONE) {
		if (open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS &&
		    open->op_delegate_type != NFS4_OPEN_DELEGATE_NONE)
			dprintk("NFSD: WARNING: refusing delegation reclaim\n");

3009 3010 3011
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
3012
	}
3013 3014
	return;
out_free:
3015
	unhash_stid(&dp->dl_stid);
3016
	nfs4_put_delegation(dp);
3017 3018 3019
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
3020 3021
}

3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
static void nfsd4_deleg_xgrade_none_ext(struct nfsd4_open *open,
					struct nfs4_delegation *dp)
{
	if (open->op_deleg_want == NFS4_SHARE_WANT_READ_DELEG &&
	    dp->dl_type == NFS4_OPEN_DELEGATE_WRITE) {
		open->op_delegate_type = NFS4_OPEN_DELEGATE_NONE_EXT;
		open->op_why_no_deleg = WND4_NOT_SUPP_DOWNGRADE;
	} else if (open->op_deleg_want == NFS4_SHARE_WANT_WRITE_DELEG &&
		   dp->dl_type == NFS4_OPEN_DELEGATE_WRITE) {
		open->op_delegate_type = NFS4_OPEN_DELEGATE_NONE_EXT;
		open->op_why_no_deleg = WND4_NOT_SUPP_UPGRADE;
	}
	/* Otherwise the client must be confused wanting a delegation
	 * it already has, therefore we don't return
	 * NFS4_OPEN_DELEGATE_NONE_EXT and reason.
	 */
}

L
Linus Torvalds 已提交
3040 3041 3042
/*
 * called with nfs4_lock_state() held.
 */
3043
__be32
L
Linus Torvalds 已提交
3044 3045
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3046
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3047
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3048 3049
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3050
	struct nfs4_ol_stateid *stp = NULL;
3051
	struct nfs4_delegation *dp = NULL;
3052
	__be32 status;
L
Linus Torvalds 已提交
3053 3054 3055 3056 3057 3058 3059 3060 3061 3062

	/*
	 * Lookup file; if found, lookup stateid and check open request,
	 * and check for delegations in the process of being recalled.
	 * If not found, create the nfs4_file struct
	 */
	fp = find_file(ino);
	if (fp) {
		if ((status = nfs4_check_open(fp, open, &stp)))
			goto out;
3063
		status = nfs4_check_deleg(cl, fp, open, &dp);
3064 3065
		if (status)
			goto out;
L
Linus Torvalds 已提交
3066
	} else {
3067
		status = nfserr_bad_stateid;
3068
		if (nfsd4_is_deleg_cur(open))
3069
			goto out;
3070
		status = nfserr_jukebox;
3071 3072 3073
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3074 3075 3076 3077 3078 3079 3080 3081
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3082
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3083 3084 3085
		if (status)
			goto out;
	} else {
3086
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3087 3088 3089
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3090
		if (status)
L
Linus Torvalds 已提交
3091
			goto out;
3092 3093
		stp = open->op_stp;
		open->op_stp = NULL;
3094
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3095
	}
3096 3097
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&open->op_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
L
Linus Torvalds 已提交
3098

3099
	if (nfsd4_has_session(&resp->cstate)) {
3100
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3101

3102 3103 3104 3105 3106 3107 3108
		if (open->op_deleg_want & NFS4_SHARE_WANT_NO_DELEG) {
			open->op_delegate_type = NFS4_OPEN_DELEGATE_NONE_EXT;
			open->op_why_no_deleg = WND4_NOT_WANTED;
			goto nodeleg;
		}
	}

L
Linus Torvalds 已提交
3109 3110 3111 3112
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3113
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3114
nodeleg:
L
Linus Torvalds 已提交
3115 3116
	status = nfs_ok;

3117
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3118
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3119
out:
3120 3121
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3122 3123
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3124

3125 3126
	if (fp)
		put_nfs4_file(fp);
3127
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3128
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3129 3130 3131 3132
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3133
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3134
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3135 3136 3137 3138 3139
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154
void nfsd4_cleanup_open_state(struct nfsd4_open *open, __be32 status)
{
	if (open->op_openowner) {
		struct nfs4_openowner *oo = open->op_openowner;

		if (!list_empty(&oo->oo_owner.so_stateids))
			list_del_init(&oo->oo_close_lru);
		if (oo->oo_flags & NFS4_OO_NEW) {
			if (status) {
				release_openowner(oo);
				open->op_openowner = NULL;
			} else
				oo->oo_flags &= ~NFS4_OO_NEW;
		}
	}
3155 3156
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3157
	if (open->op_stp)
3158
		free_generic_stateid(open->op_stp);
3159 3160
}

3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172
static __be32 lookup_clientid(clientid_t *clid, bool session, struct nfsd_net *nn, struct nfs4_client **clp)
{
	struct nfs4_client *found;

	if (STALE_CLIENTID(clid, nn))
		return nfserr_stale_clientid;
	found = find_confirmed_client(clid, session, nn);
	if (clp)
		*clp = found;
	return found ? nfs_ok : nfserr_expired;
}

3173
__be32
3174 3175
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3176 3177
{
	struct nfs4_client *clp;
3178
	__be32 status;
3179
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3180 3181 3182 3183

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3184 3185
	status = lookup_clientid(clid, cstate->minorversion, nn, &clp);
	if (status)
L
Linus Torvalds 已提交
3186 3187
		goto out;
	status = nfserr_cb_path_down;
3188
	if (!list_empty(&clp->cl_delegations)
3189
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3190 3191 3192 3193 3194 3195 3196
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3197
static void
3198
nfsd4_end_grace(struct nfsd_net *nn)
3199
{
3200
	/* do nothing if grace period already ended */
3201
	if (nn->grace_ended)
3202 3203
		return;

3204
	dprintk("NFSD: end of grace period\n");
3205
	nn->grace_ended = true;
3206
	nfsd4_record_grace_done(nn, nn->boot_time);
3207
	locks_end_grace(&nn->nfsd4_manager);
3208 3209 3210 3211 3212
	/*
	 * Now that every NFSv4 client has had the chance to recover and
	 * to see the (possibly new, possibly shorter) lease time, we
	 * can safely set the next grace time to the current lease time:
	 */
3213
	nn->nfsd4_grace = nn->nfsd4_lease;
3214 3215
}

3216
static time_t
3217
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3218 3219
{
	struct nfs4_client *clp;
3220
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3221 3222
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3223 3224 3225
	time_t cutoff = get_seconds() - nn->nfsd4_lease;
	time_t t, clientid_val = nn->nfsd4_lease;
	time_t u, test_val = nn->nfsd4_lease;
L
Linus Torvalds 已提交
3226 3227 3228 3229

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3230
	nfsd4_end_grace(nn);
3231
	INIT_LIST_HEAD(&reaplist);
3232
	spin_lock(&nn->client_lock);
3233
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3234 3235 3236 3237 3238 3239 3240
		clp = list_entry(pos, struct nfs4_client, cl_lru);
		if (time_after((unsigned long)clp->cl_time, (unsigned long)cutoff)) {
			t = clp->cl_time - cutoff;
			if (clientid_val > t)
				clientid_val = t;
			break;
		}
3241
		if (mark_client_expired_locked(clp)) {
3242 3243 3244 3245
			dprintk("NFSD: client in use (clientid %08x)\n",
				clp->cl_clientid.cl_id);
			continue;
		}
3246
		list_move(&clp->cl_lru, &reaplist);
3247
	}
3248
	spin_unlock(&nn->client_lock);
3249 3250
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3251 3252 3253 3254 3255 3256 3257
		dprintk("NFSD: purging unused client (clientid %08x)\n",
			clp->cl_clientid.cl_id);
		expire_client(clp);
	}
	spin_lock(&recall_lock);
	list_for_each_safe(pos, next, &del_recall_lru) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
3258 3259
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272
		if (time_after((unsigned long)dp->dl_time, (unsigned long)cutoff)) {
			u = dp->dl_time - cutoff;
			if (test_val > u)
				test_val = u;
			break;
		}
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	list_for_each_safe(pos, next, &reaplist) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}
3273
	test_val = nn->nfsd4_lease;
3274
	list_for_each_safe(pos, next, &nn->close_lru) {
3275 3276 3277
		oo = container_of(pos, struct nfs4_openowner, oo_close_lru);
		if (time_after((unsigned long)oo->oo_time, (unsigned long)cutoff)) {
			u = oo->oo_time - cutoff;
L
Linus Torvalds 已提交
3278 3279 3280 3281
			if (test_val > u)
				test_val = u;
			break;
		}
3282
		release_openowner(oo);
L
Linus Torvalds 已提交
3283 3284 3285 3286 3287 3288 3289
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3290 3291 3292 3293
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3294
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3295 3296
{
	time_t t;
3297 3298 3299 3300
	struct delayed_work *dwork = container_of(laundry, struct delayed_work,
						  work);
	struct nfsd_net *nn = container_of(dwork, struct nfsd_net,
					   laundromat_work);
L
Linus Torvalds 已提交
3301

3302
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3303
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3304
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3305 3306
}

3307
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3308
{
3309 3310 3311
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3312 3313 3314
}

static inline int
3315
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3316
{
3317 3318 3319
	return test_access(NFS4_SHARE_ACCESS_READ, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp) ||
		test_access(NFS4_SHARE_ACCESS_WRITE, stp);
L
Linus Torvalds 已提交
3320 3321 3322
}

static inline int
3323
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3324
{
3325 3326
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3327 3328 3329
}

static
3330
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3331
{
3332
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3333

3334 3335 3336
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3337
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3338
                goto out;
3339
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3340 3341 3342 3343 3344 3345
                goto out;
	status = nfs_ok;
out:
	return status;
}

3346
static inline __be32
3347
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3348
{
3349
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3350
		return nfs_ok;
3351
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3352
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367
		 * conflicting state; so we must wait out the grace period. */
		return nfserr_grace;
	} else if (flags & WR_STATE)
		return nfs4_share_conflict(current_fh,
				NFS4_SHARE_DENY_WRITE);
	else /* (flags & RD_STATE) && ZERO_STATEID(stateid) */
		return nfs4_share_conflict(current_fh,
				NFS4_SHARE_DENY_READ);
}

/*
 * Allow READ/WRITE during grace period on recovered state only for files
 * that are not able to provide mandatory locking.
 */
static inline int
3368
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3369
{
3370
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3371 3372
}

3373 3374 3375 3376 3377 3378
/* Returns true iff a is later than b: */
static bool stateid_generation_after(stateid_t *a, stateid_t *b)
{
	return (s32)a->si_generation - (s32)b->si_generation > 0;
}

J
J. Bruce Fields 已提交
3379
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3380
{
A
Andy Adamson 已提交
3381 3382 3383 3384
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3385
	if (has_session && in->si_generation == 0)
3386 3387 3388 3389
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3390

3391
	/* If the client sends us a stateid from the future, it's buggy: */
3392
	if (stateid_generation_after(in, ref))
3393 3394
		return nfserr_bad_stateid;
	/*
3395 3396 3397 3398 3399 3400 3401 3402
	 * However, we could see a stateid from the past, even from a
	 * non-buggy client.  For example, if the client sends a lock
	 * while some IO is outstanding, the lock may bump si_generation
	 * while the IO is still in flight.  The client could avoid that
	 * situation by waiting for responses on all the IO requests,
	 * but better performance may result in retrying IO that
	 * receives an old_stateid error if requests are rarely
	 * reordered in flight:
3403
	 */
3404
	return nfserr_old_stateid;
3405 3406
}

3407
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3408
{
3409 3410 3411
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3412

3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423
	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
	/* Client debugging aid. */
	if (!same_clid(&stateid->si_opaque.so_clid, &cl->cl_clientid)) {
		char addr_str[INET6_ADDRSTRLEN];
		rpc_ntop((struct sockaddr *)&cl->cl_addr, addr_str,
				 sizeof(addr_str));
		pr_warn_ratelimited("NFSD: client %s testing state ID "
					"with incorrect client ID\n", addr_str);
		return nfserr_bad_stateid;
	}
3424
	s = find_stateid(cl, stateid);
3425
	if (!s)
3426
		return nfserr_bad_stateid;
3427
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3428
	if (status)
3429 3430 3431 3432 3433
		return status;
	if (!(s->sc_type & (NFS4_OPEN_STID | NFS4_LOCK_STID)))
		return nfs_ok;
	ols = openlockstateid(s);
	if (ols->st_stateowner->so_is_open_owner
3434
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3435 3436
		return nfserr_bad_stateid;
	return nfs_ok;
3437 3438
}

3439 3440 3441
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3442 3443
{
	struct nfs4_client *cl;
J
J. Bruce Fields 已提交
3444
	__be32 status;
3445 3446 3447

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
J
J. Bruce Fields 已提交
3448 3449 3450
	status = lookup_clientid(&stateid->si_opaque.so_clid, sessions,
							nn, &cl);
	if (status == nfserr_stale_clientid)
3451
		return nfserr_stale_stateid;
J
J. Bruce Fields 已提交
3452 3453
	if (status)
		return status;
3454 3455 3456 3457 3458 3459
	*s = find_stateid_by_type(cl, stateid, typemask);
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;
}

L
Linus Torvalds 已提交
3460 3461 3462
/*
* Checks for stateid operations
*/
3463
__be32
3464
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3465
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3466
{
3467
	struct nfs4_stid *s;
3468
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3469
	struct nfs4_delegation *dp = NULL;
3470
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3471
	struct inode *ino = current_fh->fh_dentry->d_inode;
3472
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3473
	__be32 status;
L
Linus Torvalds 已提交
3474 3475 3476 3477

	if (filpp)
		*filpp = NULL;

3478
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3479 3480 3481
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3482
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3483

3484 3485
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3486 3487
	if (status)
		return status;
3488 3489 3490
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3491 3492
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3493
		dp = delegstateid(s);
3494 3495 3496
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3497
		if (filpp) {
3498
			*filpp = dp->dl_file->fi_deleg_file;
3499 3500 3501 3502 3503
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3504
		}
3505 3506 3507
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3508
		stp = openlockstateid(s);
3509 3510
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3511
			goto out;
3512
		if (stp->st_stateowner->so_is_open_owner
3513
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3514
			goto out;
3515 3516
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3517
			goto out;
3518 3519 3520 3521 3522 3523
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3524 3525 3526
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3527 3528 3529 3530 3531 3532
	}
	status = nfs_ok;
out:
	return status;
}

3533
static __be32
3534
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3535
{
3536
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3537 3538 3539 3540 3541
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3542 3543 3544 3545 3546 3547 3548
/*
 * Test if the stateid is valid
 */
__be32
nfsd4_test_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		   struct nfsd4_test_stateid *test_stateid)
{
3549 3550 3551 3552 3553
	struct nfsd4_test_stateid_id *stateid;
	struct nfs4_client *cl = cstate->session->se_client;

	nfs4_lock_state();
	list_for_each_entry(stateid, &test_stateid->ts_stateid_list, ts_id_list)
3554 3555
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3556 3557
	nfs4_unlock_state();

3558 3559 3560
	return nfs_ok;
}

3561 3562 3563 3564 3565
__be32
nfsd4_free_stateid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		   struct nfsd4_free_stateid *free_stateid)
{
	stateid_t *stateid = &free_stateid->fr_stateid;
J
J. Bruce Fields 已提交
3566
	struct nfs4_stid *s;
3567
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3568
	__be32 ret = nfserr_bad_stateid;
3569 3570

	nfs4_lock_state();
3571
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3572
	if (!s)
3573
		goto out;
J
J. Bruce Fields 已提交
3574 3575
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3576 3577
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3578 3579 3580 3581 3582 3583 3584 3585 3586
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
		ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
		if (ret)
			goto out;
		if (s->sc_type == NFS4_LOCK_STID)
			ret = nfsd4_free_lock_stateid(openlockstateid(s));
		else
			ret = nfserr_locks_held;
3587 3588 3589
		break;
	default:
		ret = nfserr_bad_stateid;
3590 3591 3592 3593 3594 3595
	}
out:
	nfs4_unlock_state();
	return ret;
}

3596 3597 3598 3599 3600 3601
static inline int
setlkflg (int type)
{
	return (type == NFS4_READW_LT || type == NFS4_READ_LT) ?
		RD_STATE : WR_STATE;
}
L
Linus Torvalds 已提交
3602

3603
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3604 3605 3606 3607 3608 3609 3610 3611
{
	struct svc_fh *current_fh = &cstate->current_fh;
	struct nfs4_stateowner *sop = stp->st_stateowner;
	__be32 status;

	status = nfsd4_check_seqid(cstate, sop, seqid);
	if (status)
		return status;
3612 3613 3614 3615 3616 3617 3618 3619 3620 3621
	if (stp->st_stid.sc_type == NFS4_CLOSED_STID)
		/*
		 * "Closed" stateid's exist *only* to return
		 * nfserr_replay_me from the previous step.
		 */
		return nfserr_bad_stateid;
	status = check_stateid_generation(stateid, &stp->st_stid.sc_stateid, nfsd4_has_session(cstate));
	if (status)
		return status;
	return nfs4_check_fh(current_fh, stp);
3622 3623
}

L
Linus Torvalds 已提交
3624 3625 3626
/* 
 * Checks for sequence id mutating operations. 
 */
3627
static __be32
3628
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3629
			 stateid_t *stateid, char typemask,
3630 3631
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3632
{
3633
	__be32 status;
3634
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3635

3636 3637
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3638

L
Linus Torvalds 已提交
3639
	*stpp = NULL;
3640 3641
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3642 3643
	if (status)
		return status;
3644
	*stpp = openlockstateid(s);
3645
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3646

3647 3648
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3649

3650 3651
static __be32 nfs4_preprocess_confirmed_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
						 stateid_t *stateid, struct nfs4_ol_stateid **stpp, struct nfsd_net *nn)
3652 3653 3654
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3655

3656
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3657
						NFS4_OPEN_STID, stpp, nn);
3658 3659
	if (status)
		return status;
3660
	oo = openowner((*stpp)->st_stateowner);
3661
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3662 3663
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3664 3665
}

3666
__be32
3667
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3668
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3669
{
3670
	__be32 status;
3671
	struct nfs4_openowner *oo;
3672
	struct nfs4_ol_stateid *stp;
3673
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3674 3675

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3676 3677
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3678

3679
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3680 3681
	if (status)
		return status;
L
Linus Torvalds 已提交
3682 3683 3684

	nfs4_lock_state();

3685
	status = nfs4_preprocess_seqid_op(cstate,
3686
					oc->oc_seqid, &oc->oc_req_stateid,
3687
					NFS4_OPEN_STID, &stp, nn);
3688
	if (status)
3689
		goto out;
3690
	oo = openowner(stp->st_stateowner);
3691
	status = nfserr_bad_stateid;
3692
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3693
		goto out;
3694
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3695 3696
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3697
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3698
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3699

3700
	nfsd4_client_record_create(oo->oo_owner.so_client);
3701
	status = nfs_ok;
L
Linus Torvalds 已提交
3702
out:
3703 3704
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3705 3706 3707
	return status;
}

J
J. Bruce Fields 已提交
3708
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3709
{
3710
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3711 3712
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3713
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3714
}
3715

J
J. Bruce Fields 已提交
3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729
static inline void nfs4_stateid_downgrade(struct nfs4_ol_stateid *stp, u32 to_access)
{
	switch (to_access) {
	case NFS4_SHARE_ACCESS_READ:
		nfs4_stateid_downgrade_bit(stp, NFS4_SHARE_ACCESS_WRITE);
		nfs4_stateid_downgrade_bit(stp, NFS4_SHARE_ACCESS_BOTH);
		break;
	case NFS4_SHARE_ACCESS_WRITE:
		nfs4_stateid_downgrade_bit(stp, NFS4_SHARE_ACCESS_READ);
		nfs4_stateid_downgrade_bit(stp, NFS4_SHARE_ACCESS_BOTH);
		break;
	case NFS4_SHARE_ACCESS_BOTH:
		break;
	default:
3730
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3731 3732 3733 3734
	}
}

static void
3735
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3736 3737 3738 3739
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3740
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3741 3742 3743
	}
}

3744
__be32
3745 3746
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3747
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3748
{
3749
	__be32 status;
3750
	struct nfs4_ol_stateid *stp;
3751
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3752 3753

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3754 3755
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3756

3757
	/* We don't yet support WANT bits: */
3758 3759 3760
	if (od->od_deleg_want)
		dprintk("NFSD: %s: od_deleg_want=0x%x ignored\n", __func__,
			od->od_deleg_want);
L
Linus Torvalds 已提交
3761 3762

	nfs4_lock_state();
3763
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3764
					&od->od_stateid, &stp, nn);
3765
	if (status)
L
Linus Torvalds 已提交
3766 3767
		goto out; 
	status = nfserr_inval;
3768 3769
	if (!test_access(od->od_share_access, stp)) {
		dprintk("NFSD: access not a subset current bitmap: 0x%lx, input access=%08x\n",
L
Linus Torvalds 已提交
3770 3771 3772
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3773
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3774 3775 3776 3777
		dprintk("NFSD:deny not a subset current bitmap: 0x%lx, input deny=%08x\n",
			stp->st_deny_bmap, od->od_share_deny);
		goto out;
	}
J
J. Bruce Fields 已提交
3778
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3779

3780
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3781

3782 3783
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&od->od_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
L
Linus Torvalds 已提交
3784 3785
	status = nfs_ok;
out:
3786 3787
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3788 3789 3790
	return status;
}

3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807
void nfsd4_purge_closed_stateid(struct nfs4_stateowner *so)
{
	struct nfs4_openowner *oo;
	struct nfs4_ol_stateid *s;

	if (!so->so_is_open_owner)
		return;
	oo = openowner(so);
	s = oo->oo_last_closed_stid;
	if (!s)
		return;
	if (!(oo->oo_flags & NFS4_OO_PURGE_CLOSE)) {
		/* Release the last_closed_stid on the next seqid bump: */
		oo->oo_flags |= NFS4_OO_PURGE_CLOSE;
		return;
	}
	oo->oo_flags &= ~NFS4_OO_PURGE_CLOSE;
3808 3809 3810 3811 3812 3813 3814
	release_last_closed_stateid(oo);
}

static void nfsd4_close_open_stateid(struct nfs4_ol_stateid *s)
{
	unhash_open_stateid(s);
	s->st_stid.sc_type = NFS4_CLOSED_STID;
3815 3816
}

L
Linus Torvalds 已提交
3817 3818 3819
/*
 * nfs4_unlock_state() called after encode
 */
3820
__be32
3821
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3822
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3823
{
3824
	__be32 status;
3825
	struct nfs4_openowner *oo;
3826
	struct nfs4_ol_stateid *stp;
3827 3828
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3829 3830

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3831 3832
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3833 3834

	nfs4_lock_state();
3835 3836 3837
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3838
					&stp, nn);
3839
	if (status)
L
Linus Torvalds 已提交
3840
		goto out; 
3841
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3842
	status = nfs_ok;
3843 3844
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&close->cl_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
L
Linus Torvalds 已提交
3845

3846
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3847
	release_last_closed_stateid(oo);
3848
	oo->oo_flags &= ~NFS4_OO_PURGE_CLOSE;
3849
	oo->oo_last_closed_stid = stp;
3850

3851 3852 3853 3854 3855 3856 3857 3858 3859 3860
	if (list_empty(&oo->oo_owner.so_stateids)) {
		if (cstate->minorversion) {
			release_openowner(oo);
			cstate->replay_owner = NULL;
		} else {
			/*
			 * In the 4.0 case we need to keep the owners around a
			 * little while to handle CLOSE replay.
			 */
			if (list_empty(&oo->oo_owner.so_stateids))
3861
				move_to_close_lru(oo, SVC_NET(rqstp));
3862 3863
		}
	}
L
Linus Torvalds 已提交
3864
out:
3865 3866
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3867 3868 3869
	return status;
}

3870
__be32
3871 3872
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3873
{
3874 3875
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3876
	struct nfs4_stid *s;
3877
	__be32 status;
3878
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3879

3880
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3881
		return status;
L
Linus Torvalds 已提交
3882 3883

	nfs4_lock_state();
3884 3885
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3886
	if (status)
3887
		goto out;
3888
	dp = delegstateid(s);
3889
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3890 3891 3892 3893
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3894
out:
3895 3896
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3897 3898 3899 3900 3901 3902
	return status;
}


#define LOFF_OVERFLOW(start, len)      ((u64)(len) > ~(u64)(start))

3903
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3904

B
Benny Halevy 已提交
3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919
static inline u64
end_offset(u64 start, u64 len)
{
	u64 end;

	end = start + len;
	return end >= start ? end: NFS4_MAX_UINT64;
}

/* last octet in a range */
static inline u64
last_byte_offset(u64 start, u64 len)
{
	u64 end;

3920
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3921 3922 3923 3924
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3925
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3926 3927 3928
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3929
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948
}

/*
 * TODO: Linux file offsets are _signed_ 64-bit quantities, which means that
 * we can't properly handle lock requests that go beyond the (2^63 - 1)-th
 * byte, because of sign extension problems.  Since NFSv4 calls for 64-bit
 * locking, this prevents us from being completely protocol-compliant.  The
 * real solution to this problem is to start using unsigned file offsets in
 * the VFS, but this is a very deep change!
 */
static inline void
nfs4_transform_lock_offset(struct file_lock *lock)
{
	if (lock->fl_start < 0)
		lock->fl_start = OFFSET_MAX;
	if (lock->fl_end < 0)
		lock->fl_end = OFFSET_MAX;
}

3949 3950
/* Hack!: For now, we're defining this just so we can use a pointer to it
 * as a unique cookie to identify our (NFSv4's) posix locks. */
3951
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3952
};
L
Linus Torvalds 已提交
3953 3954 3955 3956

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3957
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3958

3959
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3960 3961 3962
		lo = (struct nfs4_lockowner *) fl->fl_owner;
		deny->ld_owner.data = kmemdup(lo->lo_owner.so_owner.data,
					lo->lo_owner.so_owner.len, GFP_KERNEL);
3963 3964 3965
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3966 3967
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3968
	} else {
3969 3970 3971
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3972 3973
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3974 3975
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3976 3977
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3978 3979 3980 3981 3982 3983
		deny->ld_length = fl->fl_end - fl->fl_start + 1;        
	deny->ld_type = NFS4_READ_LT;
	if (fl->fl_type != F_RDLCK)
		deny->ld_type = NFS4_WRITE_LT;
}

J
J. Bruce Fields 已提交
3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994
static bool same_lockowner_ino(struct nfs4_lockowner *lo, struct inode *inode, clientid_t *clid, struct xdr_netobj *owner)
{
	struct nfs4_ol_stateid *lst;

	if (!same_owner_str(&lo->lo_owner, owner, clid))
		return false;
	lst = list_first_entry(&lo->lo_owner.so_stateids,
			       struct nfs4_ol_stateid, st_perstateowner);
	return lst->st_file->fi_inode == inode;
}

3995 3996
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3997
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3998
{
3999
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
4000
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4001

4002
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
4003 4004
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
4005 4006 4007 4008
	}
	return NULL;
}

4009
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
4010
{
4011 4012 4013
	struct inode *inode = open_stp->st_file->fi_inode;
	unsigned int inohash = lockowner_ino_hashval(inode,
			clp->cl_clientid.cl_id, &lo->lo_owner.so_owner);
4014
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
4015

4016
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
4017
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
4018
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
4019 4020
}

L
Linus Torvalds 已提交
4021 4022 4023
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4024
 * occurred. 
L
Linus Torvalds 已提交
4025
 *
4026
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4027 4028
 */

4029
static struct nfs4_lockowner *
4030
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4031
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4032

4033 4034
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4035
		return NULL;
4036 4037
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4038 4039
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4040 4041 4042
	lo->lo_owner.so_seqid = lock->lk_new_lock_seqid + 1;
	hash_lockowner(lo, strhashval, clp, open_stp);
	return lo;
L
Linus Torvalds 已提交
4043 4044
}

4045 4046
static struct nfs4_ol_stateid *
alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct nfs4_ol_stateid *open_stp)
L
Linus Torvalds 已提交
4047
{
4048
	struct nfs4_ol_stateid *stp;
4049
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4050

4051
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4052
	if (stp == NULL)
J
J. Bruce Fields 已提交
4053
		return NULL;
J
J. Bruce Fields 已提交
4054
	stp->st_stid.sc_type = NFS4_LOCK_STID;
4055
	list_add(&stp->st_perfile, &fp->fi_stateids);
4056 4057
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4058
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4059
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4060
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4061
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4062
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4063 4064 4065
	return stp;
}

4066
static int
L
Linus Torvalds 已提交
4067 4068
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4069
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4070 4071 4072
	     LOFF_OVERFLOW(offset, length)));
}

4073
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4074 4075 4076 4077
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4078
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4079 4080
		return;
	nfs4_file_get_access(fp, oflag);
4081
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4082 4083
}

J
J. Bruce Fields 已提交
4084
static __be32 lookup_or_create_lock_state(struct nfsd4_compound_state *cstate, struct nfs4_ol_stateid *ost, struct nfsd4_lock *lock, struct nfs4_ol_stateid **lst, bool *new)
4085 4086 4087 4088 4089 4090
{
	struct nfs4_file *fi = ost->st_file;
	struct nfs4_openowner *oo = openowner(ost->st_stateowner);
	struct nfs4_client *cl = oo->oo_owner.so_client;
	struct nfs4_lockowner *lo;
	unsigned int strhashval;
4091
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4092

4093 4094
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4095 4096 4097 4098 4099 4100 4101 4102
	if (lo) {
		if (!cstate->minorversion)
			return nfserr_bad_seqid;
		/* XXX: a lockowner always has exactly one stateid: */
		*lst = list_first_entry(&lo->lo_owner.so_stateids,
				struct nfs4_ol_stateid, st_perstateowner);
		return nfs_ok;
	}
4103
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116
			&lock->v.new.owner);
	lo = alloc_init_lock_stateowner(strhashval, cl, ost, lock);
	if (lo == NULL)
		return nfserr_jukebox;
	*lst = alloc_init_lock_stateid(lo, fi, ost);
	if (*lst == NULL) {
		release_lockowner(lo);
		return nfserr_jukebox;
	}
	*new = true;
	return nfs_ok;
}

L
Linus Torvalds 已提交
4117 4118 4119
/*
 *  LOCK operation 
 */
4120
__be32
4121
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4122
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4123
{
4124 4125
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4126
	struct nfs4_ol_stateid *lock_stp;
4127
	struct file *filp = NULL;
4128 4129
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4130
	__be32 status = 0;
4131
	bool new_state = false;
4132
	int lkflg;
4133
	int err;
4134 4135
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4136 4137 4138 4139 4140 4141 4142 4143

	dprintk("NFSD: nfsd4_lock: start=%Ld length=%Ld\n",
		(long long) lock->lk_offset,
		(long long) lock->lk_length);

	if (check_lock_length(lock->lk_offset, lock->lk_length))
		 return nfserr_inval;

4144
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4145
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4146 4147 4148 4149
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4150 4151 4152
	nfs4_lock_state();

	if (lock->lk_is_new) {
4153
		struct nfs4_ol_stateid *open_stp = NULL;
4154 4155 4156 4157 4158 4159 4160

		if (nfsd4_has_session(cstate))
			/* See rfc 5661 18.10.3: given clientid is ignored: */
			memcpy(&lock->v.new.clientid,
				&cstate->session->se_client->cl_clientid,
				sizeof(clientid_t));

L
Linus Torvalds 已提交
4161
		status = nfserr_stale_clientid;
4162
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4163 4164 4165
			goto out;

		/* validate and update open stateid and open seqid */
4166
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4167 4168
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4169
					&open_stp, nn);
4170
		if (status)
L
Linus Torvalds 已提交
4171
			goto out;
4172
		open_sop = openowner(open_stp->st_stateowner);
4173
		status = nfserr_bad_stateid;
4174
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4175 4176
						&lock->v.new.clientid))
			goto out;
4177 4178
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4179
	} else
4180
		status = nfs4_preprocess_seqid_op(cstate,
4181 4182
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4183
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4184 4185
	if (status)
		goto out;
4186
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4187

4188 4189 4190 4191 4192
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4193
	status = nfserr_grace;
4194
	if (locks_in_grace(net) && !lock->lk_reclaim)
4195 4196
		goto out;
	status = nfserr_no_grace;
4197
	if (!locks_in_grace(net) && lock->lk_reclaim)
4198 4199
		goto out;

4200 4201 4202 4203 4204 4205 4206 4207
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}

	locks_init_lock(file_lock);
L
Linus Torvalds 已提交
4208 4209 4210
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4211 4212 4213
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4214
			file_lock->fl_type = F_RDLCK;
4215
			break;
L
Linus Torvalds 已提交
4216 4217
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4218 4219 4220
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4221
			file_lock->fl_type = F_WRLCK;
4222
			break;
L
Linus Torvalds 已提交
4223 4224 4225 4226
		default:
			status = nfserr_inval;
		goto out;
	}
4227 4228 4229 4230
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245
	file_lock->fl_owner = (fl_owner_t)lock_sop;
	file_lock->fl_pid = current->tgid;
	file_lock->fl_file = filp;
	file_lock->fl_flags = FL_POSIX;
	file_lock->fl_lmops = &nfsd_posix_mng_ops;
	file_lock->fl_start = lock->lk_offset;
	file_lock->fl_end = last_byte_offset(lock->lk_offset, lock->lk_length);
	nfs4_transform_lock_offset(file_lock);

	conflock = locks_alloc_lock();
	if (!conflock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}
L
Linus Torvalds 已提交
4246

4247
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4248
	switch (-err) {
L
Linus Torvalds 已提交
4249
	case 0: /* success! */
4250 4251
		update_stateid(&lock_stp->st_stid.sc_stateid);
		memcpy(&lock->lk_resp_stateid, &lock_stp->st_stid.sc_stateid, 
L
Linus Torvalds 已提交
4252
				sizeof(stateid_t));
4253
		status = 0;
4254 4255 4256 4257
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4258
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4259
		break;
L
Linus Torvalds 已提交
4260 4261
	case (EDEADLK):
		status = nfserr_deadlock;
4262
		break;
4263
	default:
4264
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4265
		status = nfserrno(err);
4266
		break;
L
Linus Torvalds 已提交
4267 4268
	}
out:
4269
	if (status && new_state)
4270
		release_lockowner(lock_sop);
4271 4272
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4273 4274 4275 4276
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4277 4278 4279
	return status;
}

4280 4281 4282 4283 4284 4285
/*
 * The NFSv4 spec allows a client to do a LOCKT without holding an OPEN,
 * so we do a temporary open here just to get an open file to pass to
 * vfs_test_lock.  (Arguably perhaps test_lock should be done with an
 * inode operation.)
 */
4286
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4287 4288
{
	struct file *file;
4289 4290 4291 4292 4293
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4294 4295 4296
	return err;
}

L
Linus Torvalds 已提交
4297 4298 4299
/*
 * LOCKT operation
 */
4300
__be32
4301 4302
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4303 4304
{
	struct inode *inode;
4305
	struct file_lock *file_lock = NULL;
4306
	struct nfs4_lockowner *lo;
4307
	__be32 status;
4308
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4309

4310
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4311 4312 4313 4314 4315 4316 4317
		return nfserr_grace;

	if (check_lock_length(lockt->lt_offset, lockt->lt_length))
		 return nfserr_inval;

	nfs4_lock_state();

4318 4319 4320 4321 4322
	if (!nfsd4_has_session(cstate)) {
		status = lookup_clientid(&lockt->lt_clientid, false, nn, NULL);
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
4323

4324
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4325 4326
		goto out;

4327
	inode = cstate->current_fh.fh_dentry->d_inode;
4328 4329 4330 4331 4332 4333 4334
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}
	locks_init_lock(file_lock);
L
Linus Torvalds 已提交
4335 4336 4337
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4338
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4339 4340 4341
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4342
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4343 4344
		break;
		default:
4345
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4346 4347 4348 4349
			status = nfserr_inval;
		goto out;
	}

4350
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4351
	if (lo)
4352 4353 4354
		file_lock->fl_owner = (fl_owner_t)lo;
	file_lock->fl_pid = current->tgid;
	file_lock->fl_flags = FL_POSIX;
L
Linus Torvalds 已提交
4355

4356 4357
	file_lock->fl_start = lockt->lt_offset;
	file_lock->fl_end = last_byte_offset(lockt->lt_offset, lockt->lt_length);
L
Linus Torvalds 已提交
4358

4359
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4360

4361
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4362
	if (status)
4363
		goto out;
4364

4365
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4366
		status = nfserr_denied;
4367
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4368 4369 4370
	}
out:
	nfs4_unlock_state();
4371 4372
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4373 4374 4375
	return status;
}

4376
__be32
4377
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4378
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4379
{
4380
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4381
	struct file *filp = NULL;
4382
	struct file_lock *file_lock = NULL;
4383
	__be32 status;
4384
	int err;
4385 4386
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4387 4388 4389 4390 4391 4392 4393 4394 4395
	dprintk("NFSD: nfsd4_locku: start=%Ld length=%Ld\n",
		(long long) locku->lu_offset,
		(long long) locku->lu_length);

	if (check_lock_length(locku->lu_offset, locku->lu_length))
		 return nfserr_inval;

	nfs4_lock_state();
									        
4396
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4397 4398
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4399
	if (status)
L
Linus Torvalds 已提交
4400
		goto out;
4401 4402 4403 4404 4405
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}
	locks_init_lock(file_lock);
	file_lock->fl_type = F_UNLCK;
	file_lock->fl_owner = (fl_owner_t)lockowner(stp->st_stateowner);
	file_lock->fl_pid = current->tgid;
	file_lock->fl_file = filp;
	file_lock->fl_flags = FL_POSIX;
	file_lock->fl_lmops = &nfsd_posix_mng_ops;
	file_lock->fl_start = locku->lu_offset;

	file_lock->fl_end = last_byte_offset(locku->lu_offset,
						locku->lu_length);
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4424 4425 4426 4427

	/*
	*  Try to unlock the file in the VFS.
	*/
4428
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4429
	if (err) {
4430
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4431 4432 4433 4434 4435
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4436 4437
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&locku->lu_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
L
Linus Torvalds 已提交
4438 4439

out:
4440 4441
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4442 4443
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4444 4445 4446
	return status;

out_nfserr:
4447
	status = nfserrno(err);
L
Linus Torvalds 已提交
4448 4449 4450 4451 4452 4453 4454 4455 4456
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4457
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4458 4459
{
	struct file_lock **flpp;
4460
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4461 4462
	int status = 0;

4463
	lock_flocks();
L
Linus Torvalds 已提交
4464
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4465
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4466 4467
			status = 1;
			goto out;
4468
		}
L
Linus Torvalds 已提交
4469 4470
	}
out:
4471
	unlock_flocks();
L
Linus Torvalds 已提交
4472 4473 4474
	return status;
}

4475
__be32
4476 4477 4478
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4479 4480
{
	clientid_t *clid = &rlockowner->rl_clientid;
4481
	struct nfs4_stateowner *sop;
4482
	struct nfs4_lockowner *lo;
4483
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4484
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4485
	struct list_head matches;
4486
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4487
	__be32 status;
4488
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4489 4490 4491 4492 4493 4494

	dprintk("nfsd4_release_lockowner clientid: (%08x/%08x):\n",
		clid->cl_boot, clid->cl_id);

	nfs4_lock_state();

4495 4496 4497 4498
	status = lookup_clientid(clid, cstate->minorversion, nn, NULL);
	if (status)
		goto out;

4499 4500
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4501

4502
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4503 4504
		if (sop->so_is_open_owner)
			continue;
4505 4506 4507 4508 4509 4510 4511 4512
		if (!same_owner_str(sop, owner, clid))
			continue;
		list_for_each_entry(stp, &sop->so_stateids,
				st_perstateowner) {
			lo = lockowner(sop);
			if (check_for_locks(stp->st_file, lo))
				goto out;
			list_add(&lo->lo_list, &matches);
L
Linus Torvalds 已提交
4513
		}
4514 4515 4516 4517 4518
	}
	/* Clients probably won't expect us to return with some (but not all)
	 * of the lockowner state released; so don't release any until all
	 * have been checked. */
	status = nfs_ok;
N
NeilBrown 已提交
4519
	while (!list_empty(&matches)) {
4520 4521
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4522 4523
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4524 4525
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4526 4527 4528 4529 4530 4531 4532
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4533
alloc_reclaim(void)
L
Linus Torvalds 已提交
4534
{
N
NeilBrown 已提交
4535
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4536 4537
}

4538
bool
4539
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4540
{
4541
	struct nfs4_client_reclaim *crp;
4542

4543
	crp = nfsd4_find_reclaim_client(name, nn);
4544
	return (crp && crp->cr_clp);
4545 4546
}

L
Linus Torvalds 已提交
4547 4548 4549
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4550
struct nfs4_client_reclaim *
4551
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4552 4553
{
	unsigned int strhashval;
4554
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4555

N
NeilBrown 已提交
4556 4557
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4558 4559 4560
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4561
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4562
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4563
		crp->cr_clp = NULL;
4564
		nn->reclaim_str_hashtbl_size++;
4565 4566
	}
	return crp;
L
Linus Torvalds 已提交
4567 4568
}

4569
void
4570
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4571 4572 4573
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4574
	nn->reclaim_str_hashtbl_size--;
4575 4576
}

4577
void
4578
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4579 4580 4581 4582 4583
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4584 4585
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4586
			                struct nfs4_client_reclaim, cr_strhash);
4587
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4588 4589
		}
	}
4590
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4591 4592 4593 4594
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4595
struct nfs4_client_reclaim *
4596
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4597 4598 4599 4600
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4601
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4602

4603
	strhashval = clientstr_hashval(recdir);
4604
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4605
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4606 4607 4608 4609 4610 4611 4612 4613 4614
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4615
__be32
4616
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4617
{
4618 4619 4620
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4621
	clp = find_confirmed_client(clid, sessions, nn);
4622 4623 4624 4625
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4626 4627
}

B
Bryan Schumaker 已提交
4628 4629
#ifdef CONFIG_NFSD_FAULT_INJECTION

4630 4631
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
4632 4633
	if (mark_client_expired(clp))
		return 0;
4634 4635 4636 4637
	expire_client(clp);
	return 1;
}

4638 4639 4640
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
4641
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4642 4643 4644 4645 4646 4647 4648 4649
	printk(KERN_INFO "NFS Client: %s\n", buf);
	return 1;
}

static void nfsd_print_count(struct nfs4_client *clp, unsigned int count,
			     const char *type)
{
	char buf[INET6_ADDRSTRLEN];
4650
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4651 4652 4653
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

B
Bryan Schumaker 已提交
4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679
static u64 nfsd_foreach_client_lock(struct nfs4_client *clp, u64 max, void (*func)(struct nfs4_lockowner *))
{
	struct nfs4_openowner *oop;
	struct nfs4_lockowner *lop, *lo_next;
	struct nfs4_ol_stateid *stp, *st_next;
	u64 count = 0;

	list_for_each_entry(oop, &clp->cl_openowners, oo_perclient) {
		list_for_each_entry_safe(stp, st_next, &oop->oo_owner.so_stateids, st_perstateowner) {
			list_for_each_entry_safe(lop, lo_next, &stp->st_lockowners, lo_perstateid) {
				if (func)
					func(lop);
				if (++count == max)
					return count;
			}
		}
	}

	return count;
}

u64 nfsd_forget_client_locks(struct nfs4_client *clp, u64 max)
{
	return nfsd_foreach_client_lock(clp, max, release_lockowner);
}

4680 4681 4682 4683 4684 4685 4686
u64 nfsd_print_client_locks(struct nfs4_client *clp, u64 max)
{
	u64 count = nfsd_foreach_client_lock(clp, max, NULL);
	nfsd_print_count(clp, count, "locked files");
	return count;
}

4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706
static u64 nfsd_foreach_client_open(struct nfs4_client *clp, u64 max, void (*func)(struct nfs4_openowner *))
{
	struct nfs4_openowner *oop, *next;
	u64 count = 0;

	list_for_each_entry_safe(oop, next, &clp->cl_openowners, oo_perclient) {
		if (func)
			func(oop);
		if (++count == max)
			break;
	}

	return count;
}

u64 nfsd_forget_client_openowners(struct nfs4_client *clp, u64 max)
{
	return nfsd_foreach_client_open(clp, max, release_openowner);
}

4707 4708 4709 4710 4711 4712 4713
u64 nfsd_print_client_openowners(struct nfs4_client *clp, u64 max)
{
	u64 count = nfsd_foreach_client_open(clp, max, NULL);
	nfsd_print_count(clp, count, "open files");
	return count;
}

4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759
static u64 nfsd_find_all_delegations(struct nfs4_client *clp, u64 max,
				     struct list_head *victims)
{
	struct nfs4_delegation *dp, *next;
	u64 count = 0;

	list_for_each_entry_safe(dp, next, &clp->cl_delegations, dl_perclnt) {
		if (victims)
			list_move(&dp->dl_recall_lru, victims);
		if (++count == max)
			break;
	}
	return count;
}

u64 nfsd_forget_client_delegations(struct nfs4_client *clp, u64 max)
{
	struct nfs4_delegation *dp, *next;
	LIST_HEAD(victims);
	u64 count;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, &victims);
	spin_unlock(&recall_lock);

	list_for_each_entry_safe(dp, next, &victims, dl_recall_lru)
		unhash_delegation(dp);

	return count;
}

u64 nfsd_recall_client_delegations(struct nfs4_client *clp, u64 max)
{
	struct nfs4_delegation *dp, *next;
	LIST_HEAD(victims);
	u64 count;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, &victims);
	list_for_each_entry_safe(dp, next, &victims, dl_recall_lru)
		nfsd_break_one_deleg(dp);
	spin_unlock(&recall_lock);

	return count;
}

4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771
u64 nfsd_print_client_delegations(struct nfs4_client *clp, u64 max)
{
	u64 count = 0;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, NULL);
	spin_unlock(&recall_lock);

	nfsd_print_count(clp, count, "delegations");
	return count;
}

4772
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4773 4774
{
	struct nfs4_client *clp, *next;
4775
	u64 count = 0;
4776
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4777

4778 4779 4780
	if (!nfsd_netns_ready(nn))
		return 0;

4781
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4782 4783
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4784 4785 4786
			break;
	}

4787 4788 4789
	return count;
}

4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804
struct nfs4_client *nfsd_find_client(struct sockaddr_storage *addr, size_t addr_size)
{
	struct nfs4_client *clp;
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);

	if (!nfsd_netns_ready(nn))
		return NULL;

	list_for_each_entry(clp, &nn->client_lru, cl_lru) {
		if (memcmp(&clp->cl_addr, addr, addr_size) == 0)
			return clp;
	}
	return NULL;
}

B
Bryan Schumaker 已提交
4805 4806
#endif /* CONFIG_NFSD_FAULT_INJECTION */

4807
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4808

4809
void
4810
nfs4_state_init(void)
L
Linus Torvalds 已提交
4811 4812
{
	INIT_LIST_HEAD(&del_recall_lru);
4813 4814
}

4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835
/*
 * Since the lifetime of a delegation isn't limited to that of an open, a
 * client may quite reasonably hang on to a delegation as long as it has
 * the inode cached.  This becomes an obvious problem the first time a
 * client's inode cache approaches the size of the server's total memory.
 *
 * For now we avoid this problem by imposing a hard limit on the number
 * of delegations, which varies according to the server's memory size.
 */
static void
set_max_delegations(void)
{
	/*
	 * Allow at most 4 delegations per megabyte of RAM.  Quick
	 * estimates suggest that in the worst case (where every delegation
	 * is for a different inode), a delegation could take about 1.5K,
	 * giving a worst case usage of about 6% of memory.
	 */
	max_delegations = nr_free_buffer_pages() >> (20 - 2 - PAGE_SHIFT);
}

4836
static int nfs4_state_create_net(struct net *net)
4837 4838 4839 4840 4841 4842 4843
{
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
	int i;

	nn->conf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->conf_id_hashtbl)
4844
		goto err;
4845 4846 4847 4848
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4849 4850 4851 4852
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4853 4854 4855 4856
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4857 4858 4859 4860
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4861

4862
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4863
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4864
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4865
	}
4866 4867
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4868 4869
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4870 4871
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4872
	nn->conf_name_tree = RB_ROOT;
4873
	nn->unconf_name_tree = RB_ROOT;
4874
	INIT_LIST_HEAD(&nn->client_lru);
4875
	INIT_LIST_HEAD(&nn->close_lru);
4876
	spin_lock_init(&nn->client_lock);
4877

4878
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4879
	get_net(net);
4880

4881
	return 0;
4882

4883 4884
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4885 4886
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4887 4888
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4889 4890
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4891 4892
err:
	return -ENOMEM;
4893 4894 4895
}

static void
4896
nfs4_state_destroy_net(struct net *net)
4897 4898 4899 4900
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4901
	struct rb_node *node, *tmp;
4902 4903 4904 4905 4906 4907 4908

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
		while (!list_empty(&nn->conf_id_hashtbl[i])) {
			clp = list_entry(nn->conf_id_hashtbl[i].next, struct nfs4_client, cl_idhash);
			destroy_client(clp);
		}
	}
4909 4910 4911 4912 4913 4914 4915 4916 4917 4918

	node = rb_first(&nn->unconf_name_tree);
	while (node != NULL) {
		tmp = node;
		node = rb_next(tmp);
		clp = rb_entry(tmp, struct nfs4_client, cl_namenode);
		rb_erase(tmp, &nn->unconf_name_tree);
		destroy_client(clp);
	}

4919
	kfree(nn->sessionid_hashtbl);
4920
	kfree(nn->lockowner_ino_hashtbl);
4921
	kfree(nn->ownerstr_hashtbl);
4922
	kfree(nn->unconf_id_hashtbl);
4923
	kfree(nn->conf_id_hashtbl);
4924
	put_net(net);
4925 4926
}

4927
int
4928
nfs4_state_start_net(struct net *net)
4929
{
4930
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4931 4932
	int ret;

4933
	ret = nfs4_state_create_net(net);
4934 4935
	if (ret)
		return ret;
4936
	nfsd4_client_tracking_init(net);
4937
	nn->boot_time = get_seconds();
4938
	locks_start_grace(net, &nn->nfsd4_manager);
4939
	nn->grace_ended = false;
4940
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4941 4942
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4943 4944 4945 4946 4947 4948 4949 4950 4951 4952
	return 0;
}

/* initialization to perform when the nfsd service is started: */

int
nfs4_state_start(void)
{
	int ret;

4953
	ret = set_callback_cred();
4954 4955
	if (ret)
		return -ENOMEM;
4956
	laundry_wq = create_singlethread_workqueue("nfsd4");
4957 4958 4959 4960
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4961 4962 4963
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4964

4965
	set_max_delegations();
4966

4967
	return 0;
4968

4969 4970
out_free_laundry:
	destroy_workqueue(laundry_wq);
4971
out_recovery:
4972
	return ret;
L
Linus Torvalds 已提交
4973 4974
}

4975
/* should be called with the state lock held */
4976
void
4977
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4978 4979 4980
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4981
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4982

4983 4984
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4985

L
Linus Torvalds 已提交
4986 4987 4988 4989
	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
	list_for_each_safe(pos, next, &del_recall_lru) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
4990 4991
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4992 4993 4994 4995 4996 4997 4998 4999
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	list_for_each_safe(pos, next, &reaplist) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}

5000
	nfsd4_client_tracking_exit(net);
5001
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
5002 5003 5004 5005 5006
}

void
nfs4_state_shutdown(void)
{
5007
	destroy_workqueue(laundry_wq);
5008
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
5009
}
5010 5011 5012 5013

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5014 5015
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
5016 5017 5018 5019 5020
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5021 5022 5023 5024 5025 5026 5027 5028 5029 5030
	if (cstate->minorversion) {
		memcpy(&cstate->current_stateid, stateid, sizeof(stateid_t));
		SET_STATE_ID(cstate, CURRENT_STATE_ID_FLAG);
	}
}

void
clear_current_stateid(struct nfsd4_compound_state *cstate)
{
	CLEAR_STATE_ID(cstate, CURRENT_STATE_ID_FLAG);
5031 5032
}

5033 5034 5035
/*
 * functions to set current state id
 */
5036 5037 5038 5039 5040 5041
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5042 5043 5044 5045 5046 5047
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062
void
nfsd4_set_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	put_stateid(cstate, &close->cl_stateid);
}

void
nfsd4_set_lockstateid(struct nfsd4_compound_state *cstate, struct nfsd4_lock *lock)
{
	put_stateid(cstate, &lock->lk_resp_stateid);
}

/*
 * functions to consume current state id
 */
5063

5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075
void
nfsd4_get_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	get_stateid(cstate, &odp->od_stateid);
}

void
nfsd4_get_delegreturnstateid(struct nfsd4_compound_state *cstate, struct nfsd4_delegreturn *drp)
{
	get_stateid(cstate, &drp->dr_stateid);
}

5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087
void
nfsd4_get_freestateid(struct nfsd4_compound_state *cstate, struct nfsd4_free_stateid *fsp)
{
	get_stateid(cstate, &fsp->fr_stateid);
}

void
nfsd4_get_setattrstateid(struct nfsd4_compound_state *cstate, struct nfsd4_setattr *setattr)
{
	get_stateid(cstate, &setattr->sa_stateid);
}

5088 5089 5090 5091 5092 5093 5094
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5095
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5096
{
5097
	get_stateid(cstate, &locku->lu_stateid);
5098
}
5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110

void
nfsd4_get_readstateid(struct nfsd4_compound_state *cstate, struct nfsd4_read *read)
{
	get_stateid(cstate, &read->rd_stateid);
}

void
nfsd4_get_writestateid(struct nfsd4_compound_state *cstate, struct nfsd4_write *write)
{
	get_stateid(cstate, &write->wr_stateid);
}