nfs4state.c 132.1 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 98 99 100 101 102 103 104 105 106 107 108 109
static void free_session(struct kref *);

/* Must be called under the client_lock */
static void nfsd4_put_session_locked(struct nfsd4_session *ses)
{
	kref_put(&ses->se_ref, free_session);
}

static void nfsd4_get_session(struct nfsd4_session *ses)
{
	kref_get(&ses->se_ref);
}

L
Linus Torvalds 已提交
110 111 112
void
nfs4_unlock_state(void)
{
I
Ingo Molnar 已提交
113
	mutex_unlock(&client_mutex);
L
Linus Torvalds 已提交
114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
}

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;

131 132 133 134 135
static void nfsd4_free_file(struct nfs4_file *f)
{
	kmem_cache_free(file_slab, f);
}

136 137 138
static inline void
put_nfs4_file(struct nfs4_file *fi)
{
139 140 141 142
	if (atomic_dec_and_lock(&fi->fi_ref, &recall_lock)) {
		list_del(&fi->fi_hash);
		spin_unlock(&recall_lock);
		iput(fi->fi_inode);
143
		nfsd4_free_file(fi);
144
	}
145 146 147 148 149
}

static inline void
get_nfs4_file(struct nfs4_file *fi)
{
150
	atomic_inc(&fi->fi_ref);
151 152
}

153
static int num_delegations;
154
unsigned long max_delegations;
155 156 157 158 159

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

160 161 162 163
/* 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)
164

165
static unsigned int ownerstr_hashval(u32 clientid, struct xdr_netobj *ownername)
166 167 168 169 170
{
	unsigned int ret;

	ret = opaque_hashval(ownername->data, ownername->len);
	ret += clientid;
171
	return ret & OWNER_HASH_MASK;
172
}
173 174 175 176

/* hash table for nfs4_file */
#define FILE_HASH_BITS                   8
#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
S
Shan Wei 已提交
177

178 179 180 181 182 183
static unsigned int file_hashval(struct inode *ino)
{
	/* XXX: why are we hashing on inode pointer, anyway? */
	return hash_ptr(ino, FILE_HASH_BITS);
}

184 185
static struct list_head file_hashtbl[FILE_HASH_SIZE];

186
static void __nfs4_file_get_access(struct nfs4_file *fp, int oflag)
187
{
188
	WARN_ON_ONCE(!(fp->fi_fds[oflag] || fp->fi_fds[O_RDWR]));
189 190 191
	atomic_inc(&fp->fi_access[oflag]);
}

192 193 194 195 196 197 198 199 200 201
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)
202 203 204 205 206 207 208
{
	if (fp->fi_fds[oflag]) {
		fput(fp->fi_fds[oflag]);
		fp->fi_fds[oflag] = NULL;
	}
}

209
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
210 211 212
{
	if (atomic_dec_and_test(&fp->fi_access[oflag])) {
		nfs4_file_put_fd(fp, oflag);
213 214 215 216 217 218 219 220
		/*
		 * 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);
221 222 223
	}
}

224 225 226 227 228 229 230 231 232
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
233 234
static struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl, struct
kmem_cache *slab)
235
{
J
J. Bruce Fields 已提交
236 237 238
	struct idr *stateids = &cl->cl_stateids;
	static int min_stateid = 0;
	struct nfs4_stid *stid;
J
J. Bruce Fields 已提交
239
	int new_id;
240

J
J. Bruce Fields 已提交
241 242 243 244
	stid = kmem_cache_alloc(slab, GFP_KERNEL);
	if (!stid)
		return NULL;

T
Tejun Heo 已提交
245 246
	new_id = idr_alloc(stateids, stid, min_stateid, 0, GFP_KERNEL);
	if (new_id < 0)
J
J. Bruce Fields 已提交
247
		goto out_free;
248
	stid->sc_client = cl;
J
J. Bruce Fields 已提交
249 250 251
	stid->sc_type = 0;
	stid->sc_stateid.si_opaque.so_id = new_id;
	stid->sc_stateid.si_opaque.so_clid = cl->cl_clientid;
252
	/* Will be incremented before return to client: */
J
J. Bruce Fields 已提交
253
	stid->sc_stateid.si_generation = 0;
254 255

	/*
J
J. Bruce Fields 已提交
256 257 258 259 260 261 262
	 * 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):
263
	 */
J
J. Bruce Fields 已提交
264 265 266 267 268 269 270 271

	min_stateid = new_id+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return stid;
out_free:
	kfree(stid);
	return NULL;
272 273
}

274 275 276 277 278
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
279
static struct nfs4_delegation *
280
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
281 282 283 284 285
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
286 287 288 289 290 291 292
	/*
	 * 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;
293 294
	if (fp->fi_had_conflict)
		return NULL;
295
	if (num_delegations > max_delegations)
296
		return NULL;
297
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
298
	if (dp == NULL)
L
Linus Torvalds 已提交
299
		return dp;
J
J. Bruce Fields 已提交
300
	dp->dl_stid.sc_type = NFS4_DELEG_STID;
301 302
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
303 304
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
305 306
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
307
	num_delegations++;
308 309
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
310
	INIT_LIST_HEAD(&dp->dl_recall_lru);
311
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
312 313
	dp->dl_file = fp;
	dp->dl_type = type;
314
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
315 316
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
317
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
318 319 320
	return dp;
}

F
Fengguang Wu 已提交
321
static void free_stid(struct nfs4_stid *s, struct kmem_cache *slab)
J
J. Bruce Fields 已提交
322 323 324 325 326 327 328
{
	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
329 330 331 332 333
void
nfs4_put_delegation(struct nfs4_delegation *dp)
{
	if (atomic_dec_and_test(&dp->dl_count)) {
		dprintk("NFSD: freeing dp %p\n",dp);
334
		put_nfs4_file(dp->dl_file);
J
J. Bruce Fields 已提交
335
		free_stid(&dp->dl_stid, deleg_slab);
336
		num_delegations--;
L
Linus Torvalds 已提交
337 338 339
	}
}

340
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
341
{
342 343 344
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
345
		fput(fp->fi_deleg_file);
346 347
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
348 349
}

J
J. Bruce Fields 已提交
350 351
static void unhash_stid(struct nfs4_stid *s)
{
J
J. Bruce Fields 已提交
352
	s->sc_type = 0;
J
J. Bruce Fields 已提交
353 354
}

L
Linus Torvalds 已提交
355 356 357 358
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
359
	unhash_stid(&dp->dl_stid);
360
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
361
	spin_lock(&recall_lock);
362
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
363 364
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
365
	nfs4_put_deleg_lease(dp->dl_file);
L
Linus Torvalds 已提交
366 367 368 369 370 371 372
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

373 374 375 376 377 378 379 380 381 382
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;
}

383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
/*
 * 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.
 */
401 402
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
403
	int i;
404
	unsigned int access = 0;
405 406 407

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
408
			access |= i;
409
	}
410
	return access;
411 412
}

413
static bool
414
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
415 416
	unsigned int access, deny;

417 418
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
419
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
420 421
		return false;
	return true;
422 423
}

424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
/* 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);
}

445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463
/* 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);
464 465 466 467
}

static int nfs4_access_to_omode(u32 access)
{
468
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
469 470 471 472 473 474 475
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
476 477
	WARN_ON_ONCE(1);
	return O_RDONLY;
478 479
}

480 481 482 483 484 485 486 487 488 489 490 491 492 493
/* 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);
	}
}

494
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
495 496 497 498 499
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

500
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
501
{
502
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
503
	put_nfs4_file(stp->st_file);
504 505 506
	stp->st_file = NULL;
}

507
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
508
{
J
J. Bruce Fields 已提交
509
	free_stid(&stp->st_stid, stateid_slab);
510 511
}

512
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
513 514 515 516
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
517
	unhash_stid(&stp->st_stid);
518 519
	file = find_any_file(stp->st_file);
	if (file)
520
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
521
	close_generic_stateid(stp);
522 523 524
	free_generic_stateid(stp);
}

525
static void unhash_lockowner(struct nfs4_lockowner *lo)
526
{
527
	struct nfs4_ol_stateid *stp;
528

529 530
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
531
	list_del(&lo->lo_owner_ino_hash);
532 533
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
534
				struct nfs4_ol_stateid, st_perstateowner);
535 536 537 538
		release_lock_stateid(stp);
	}
}

539
static void release_lockowner(struct nfs4_lockowner *lo)
540
{
541 542
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
543 544 545
}

static void
546
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
547
{
548
	struct nfs4_lockowner *lo;
549 550

	while (!list_empty(&open_stp->st_lockowners)) {
551 552 553
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
554 555 556
	}
}

557
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
558 559 560
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
561 562 563 564 565 566
	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
567
	unhash_stid(&stp->st_stid);
568 569 570
	free_generic_stateid(stp);
}

571
static void unhash_openowner(struct nfs4_openowner *oo)
572
{
573
	struct nfs4_ol_stateid *stp;
574

575 576 577 578
	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,
579
				struct nfs4_ol_stateid, st_perstateowner);
580
		release_open_stateid(stp);
581 582 583
	}
}

584 585 586 587 588
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
589
		unhash_stid(&s->st_stid);
590 591 592 593 594
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

595
static void release_openowner(struct nfs4_openowner *oo)
596
{
597 598
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
599
	release_last_closed_stateid(oo);
600
	nfs4_free_openowner(oo);
601 602
}

M
Marc Eshel 已提交
603 604 605 606 607 608 609 610
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

611
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
612 613 614 615 616 617
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]);
}
618 619 620 621 622 623 624
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

M
Marc Eshel 已提交
625

A
Andy Adamson 已提交
626 627 628 629 630 631 632 633 634 635 636 637 638
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;
}

/*
639 640 641 642 643 644 645 646 647 648 649 650 651
 * 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)

652 653 654 655 656 657 658 659 660
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]);
}

661
/*
662 663 664 665 666 667 668 669
 * 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;
}

670
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
671
{
672 673
	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
674

675 676
	return size;
}
A
Andy Adamson 已提交
677

678 679
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
680
 * re-negotiate active sessions and reduce their slot usage to make
681
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
682
 */
683
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
684
{
685
	int avail;
A
Andy Adamson 已提交
686

687
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
688

689
	spin_lock(&nfsd_drc_lock);
690 691
	avail = min((unsigned long)NFSD_MAX_MEM_PER_SESSION,
		    nfsd_drc_max_mem - nfsd_drc_mem_used);
692 693 694
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
695

696 697
	return num;
}
A
Andy Adamson 已提交
698

699 700
static void nfsd4_put_drc_mem(int slotsize, int num)
{
701
	spin_lock(&nfsd_drc_lock);
702
	nfsd_drc_mem_used -= slotsize * num;
703
	spin_unlock(&nfsd_drc_lock);
704
}
A
Andy Adamson 已提交
705

706
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
707 708 709
{
	struct nfsd4_session *new;
	int mem, i;
710

711 712 713
	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
714

715 716 717
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
718
	/* allocate each struct nfsd4_slot and data cache in one piece */
719 720 721 722
	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])
723 724
			goto out_free;
	}
725 726 727 728 729 730
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
731 732
}

733 734 735 736
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
737
{
738
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
739

740
	new->maxreqs = numslots;
741 742
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
743 744 745 746
	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
747

748 749 750 751 752
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
753

754 755 756 757
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
758

759 760 761 762 763
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
764
	nfsd4_probe_callback(clp);
765
	spin_unlock(&clp->cl_lock);
766
}
A
Andy Adamson 已提交
767

768
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
769 770
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
771

772 773
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
774
		return NULL;
775 776
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
777
	conn->cn_flags = flags;
778 779 780
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
781

782 783 784 785
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
786 787
}

788
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
789
{
790
	struct nfs4_client *clp = ses->se_client;
791

792
	spin_lock(&clp->cl_lock);
793
	__nfsd4_hash_conn(conn, ses);
794
	spin_unlock(&clp->cl_lock);
795 796
}

797
static int nfsd4_register_conn(struct nfsd4_conn *conn)
798
{
799
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
800
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
801 802
}

803
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
804
{
805
	int ret;
A
Andy Adamson 已提交
806

807
	nfsd4_hash_conn(conn, ses);
808 809 810 811
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
812
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
813 814 815
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
816
}
A
Andy Adamson 已提交
817

818
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
819 820 821
{
	u32 dir = NFS4_CDFC4_FORE;

822
	if (cses->flags & SESSION4_BACK_CHAN)
823
		dir |= NFS4_CDFC4_BACK;
824
	return alloc_conn(rqstp, dir);
825 826 827
}

/* must be called under client_lock */
828
static void nfsd4_del_conns(struct nfsd4_session *s)
829
{
830 831
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
832

833 834 835 836 837
	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);
838

839 840
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
841

842 843 844
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
845
}
A
Andy Adamson 已提交
846

847 848 849 850 851 852 853
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);
}

854
static void free_session(struct kref *kref)
855 856
{
	struct nfsd4_session *ses;
857
	struct nfsd_net *nn;
858 859

	ses = container_of(kref, struct nfsd4_session, se_ref);
860 861 862
	nn = net_generic(ses->se_client->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
863
	nfsd4_del_conns(ses);
864
	__free_session(ses);
865 866
}

867
static void nfsd4_put_session(struct nfsd4_session *ses)
868
{
869 870 871
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
872
	nfsd4_put_session_locked(ses);
873
	spin_unlock(&nn->client_lock);
874 875
}

876 877
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan,
					   struct nfsd_net *nn)
878 879 880 881 882 883 884 885 886 887 888 889
{
	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);
890 891
	if (numslots < 1)
		return NULL;
892

893
	new = __alloc_session(slotsize, numslots);
894
	if (!new) {
895
		nfsd4_put_drc_mem(slotsize, numslots);
896
		return NULL;
897
	}
898
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize, nn);
899 900
	return new;
}
901

902
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
903 904
{
	int idx;
905
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
906

A
Andy Adamson 已提交
907 908 909
	new->se_client = clp;
	gen_sessionid(new);

910 911
	INIT_LIST_HEAD(&new->se_conns);

912
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
913
	new->se_flags = cses->flags;
914
	new->se_cb_prog = cses->callback_prog;
915
	new->se_cb_sec = cses->cb_sec;
A
Andy Adamson 已提交
916
	kref_init(&new->se_ref);
917
	idx = hash_sessionid(&new->se_sessionid);
918
	spin_lock(&nn->client_lock);
919
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
920
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
921
	list_add(&new->se_perclnt, &clp->cl_sessions);
922
	spin_unlock(&clp->cl_lock);
923
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
924

925
	if (cses->flags & SESSION4_BACK_CHAN) {
926
		struct sockaddr *sa = svc_addr(rqstp);
927 928 929 930 931 932 933
		/*
		 * 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:
		 */
934 935 936
		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
937 938
}

939
/* caller must hold client_lock */
M
Marc Eshel 已提交
940
static struct nfsd4_session *
941
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
942 943 944
{
	struct nfsd4_session *elem;
	int idx;
945
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
946 947 948 949

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
950
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
951 952 953 954 955 956 957 958 959 960
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

961
/* caller must hold client_lock */
A
Andy Adamson 已提交
962
static void
M
Marc Eshel 已提交
963
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
964 965
{
	list_del(&ses->se_hash);
966
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
967
	list_del(&ses->se_perclnt);
968
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
969 970
}

971
/* must be called under the client_lock */
L
Linus Torvalds 已提交
972
static inline void
973
renew_client_locked(struct nfs4_client *clp)
L
Linus Torvalds 已提交
974
{
975 976
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

B
Benny Halevy 已提交
977
	if (is_client_expired(clp)) {
978 979
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
B
Benny Halevy 已提交
980 981 982 983 984 985
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

L
Linus Torvalds 已提交
986 987 988
	dprintk("renewing client (clientid %08x/%08x)\n", 
			clp->cl_clientid.cl_boot, 
			clp->cl_clientid.cl_id);
989
	list_move_tail(&clp->cl_lru, &nn->client_lru);
L
Linus Torvalds 已提交
990 991 992
	clp->cl_time = get_seconds();
}

993 994 995
static inline void
renew_client(struct nfs4_client *clp)
{
996 997 998
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
999
	renew_client_locked(clp);
1000
	spin_unlock(&nn->client_lock);
1001 1002
}

L
Linus Torvalds 已提交
1003 1004
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1005
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1006
{
1007
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1008
		return 0;
A
Andy Adamson 已提交
1009
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1010
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015 1016 1017 1018
	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.
 */
1019
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1020 1021 1022
{
	struct nfs4_client *clp;

1023 1024 1025
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1026
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1027 1028 1029
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1030
	}
1031
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1032 1033 1034 1035 1036 1037
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1038
	struct nfsd_net __maybe_unused *nn = net_generic(clp->net, nfsd_net_id);
1039 1040

	lockdep_assert_held(&nn->client_lock);
1041 1042 1043 1044 1045
	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);
1046
		nfsd4_put_session_locked(ses);
1047
	}
1048
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1049
	kfree(clp->cl_name.data);
M
majianpeng 已提交
1050
	idr_destroy(&clp->cl_stateids);
L
Linus Torvalds 已提交
1051 1052 1053
	kfree(clp);
}

1054 1055 1056 1057
void
release_session_client(struct nfsd4_session *session)
{
	struct nfs4_client *clp = session->se_client;
1058
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
1059

1060
	nfsd4_put_session(session);
1061
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
1062
		return;
1063 1064 1065 1066 1067
	/*
	 * At this point we also know all sessions have refcnt 1,
	 * so free_client will delete them all if necessary:
	 */
	if (is_client_expired(clp))
1068
		free_client(clp);
1069
	else
1070
		renew_client_locked(clp);
1071
	spin_unlock(&nn->client_lock);
1072 1073
}

B
Benny Halevy 已提交
1074 1075 1076 1077
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1078 1079
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1080
	mark_client_expired(clp);
B
Benny Halevy 已提交
1081
	list_del(&clp->cl_lru);
1082
	spin_lock(&clp->cl_lock);
1083 1084
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1085
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1086 1087
}

L
Linus Torvalds 已提交
1088
static void
1089
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1090
{
1091
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1092 1093
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1094
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1095 1096 1097

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1098 1099 1100
	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
1101 1102 1103 1104 1105 1106 1107
		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);
	}
1108
	while (!list_empty(&clp->cl_openowners)) {
1109 1110
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1111
	}
1112
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1113 1114
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1115
	list_del(&clp->cl_idhash);
1116
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1117
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1118
	else
1119
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1120
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1121
	unhash_client_locked(clp);
1122 1123
	if (atomic_read(&clp->cl_refcount) == 0)
		free_client(clp);
1124
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1125 1126
}

1127 1128 1129 1130 1131 1132
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1133 1134 1135 1136
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
1137 1138
}

1139 1140
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1141 1142 1143 1144
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1145
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1146
{
1147 1148 1149 1150 1151 1152 1153
	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;
1154
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1155 1156 1157 1158
	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);
1159
	return 0;
L
Linus Torvalds 已提交
1160 1161
}

1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
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);
}

1173
static int same_name(const char *n1, const char *n2)
1174
{
N
NeilBrown 已提交
1175
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1176 1177 1178
}

static int
1179 1180 1181
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1182 1183 1184
}

static int
1185 1186 1187
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
1188 1189
}

1190 1191 1192 1193 1194 1195 1196
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++)
1197
		if (!gid_eq(GROUP_AT(g1, i), GROUP_AT(g2, i)))
1198 1199 1200 1201
			return false;
	return true;
}

1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
/*
 * 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);
}


1218
static bool
1219 1220
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1221
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1222 1223
		|| (!uid_eq(cr1->cr_uid, cr2->cr_uid))
		|| (!gid_eq(cr1->cr_gid, cr2->cr_gid))
1224 1225 1226 1227 1228 1229
		|| !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;
1230
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1231 1232
}

1233
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1234 1235 1236
{
	static u32 current_clientid = 1;

1237
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1238 1239 1240
	clp->cl_clientid.cl_id = current_clientid++; 
}

1241 1242
static void gen_confirm(struct nfs4_client *clp)
{
1243
	__be32 verf[2];
1244
	static u32 i;
L
Linus Torvalds 已提交
1245

1246 1247 1248
	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
1249 1250
}

1251
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1252
{
J
J. Bruce Fields 已提交
1253 1254 1255 1256 1257 1258
	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
1259 1260
}

1261
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1262 1263
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1264

1265
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1266 1267
	if (!s)
		return NULL;
1268
	if (typemask & s->sc_type)
1269 1270 1271 1272
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1273
static struct nfs4_client *create_client(struct xdr_netobj name,
1274 1275 1276 1277
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1278
	int ret;
1279
	struct net *net = SVC_NET(rqstp);
1280
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1281 1282 1283 1284 1285

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

1286
	INIT_LIST_HEAD(&clp->cl_sessions);
1287 1288
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1289
		spin_lock(&nn->client_lock);
1290
		free_client(clp);
1291
		spin_unlock(&nn->client_lock);
1292
		return NULL;
1293
	}
1294
	idr_init(&clp->cl_stateids);
1295
	atomic_set(&clp->cl_refcount, 0);
1296
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1297 1298 1299 1300
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1301
	INIT_LIST_HEAD(&clp->cl_callbacks);
1302
	spin_lock_init(&clp->cl_lock);
1303
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1304
	clp->cl_time = get_seconds();
1305 1306 1307 1308 1309
	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);
1310
	clp->cl_cb_session = NULL;
1311
	clp->net = net;
1312 1313 1314
	return clp;
}

1315
static void
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
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
1357 1358
{
	unsigned int idhashval;
1359
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1360

1361
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1362
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1363
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1364
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1365
	renew_client(clp);
L
Linus Torvalds 已提交
1366 1367
}

1368
static void
L
Linus Torvalds 已提交
1369 1370 1371
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1372
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1373 1374

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1375
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1376
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1377
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1378
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1379 1380 1381 1382
	renew_client(clp);
}

static struct nfs4_client *
J
J. Bruce Fields 已提交
1383
find_client_in_id_table(struct list_head *tbl, clientid_t *clid, bool sessions)
L
Linus Torvalds 已提交
1384 1385 1386 1387
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

J
J. Bruce Fields 已提交
1388
	list_for_each_entry(clp, &tbl[idhashval], cl_idhash) {
1389
		if (same_clid(&clp->cl_clientid, clid)) {
1390 1391
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1392
			renew_client(clp);
L
Linus Torvalds 已提交
1393
			return clp;
1394
		}
L
Linus Torvalds 已提交
1395 1396 1397 1398
	}
	return NULL;
}

J
J. Bruce Fields 已提交
1399 1400 1401 1402 1403 1404 1405 1406
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
1407
static struct nfs4_client *
1408
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1409
{
J
J. Bruce Fields 已提交
1410
	struct list_head *tbl = nn->unconf_id_hashtbl;
L
Linus Torvalds 已提交
1411

J
J. Bruce Fields 已提交
1412
	return find_client_in_id_table(tbl, clid, sessions);
L
Linus Torvalds 已提交
1413 1414
}

1415
static bool clp_used_exchangeid(struct nfs4_client *clp)
1416
{
1417
	return clp->cl_exchange_flags != 0;
1418
} 
1419

1420
static struct nfs4_client *
1421
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1422
{
1423
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1424 1425 1426
}

static struct nfs4_client *
1427
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1428
{
1429
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1430 1431
}

1432
static void
1433
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1434
{
1435
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1436 1437
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447
	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
1448 1449
		goto out_err;

1450
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1451
					    se->se_callback_addr_len,
1452 1453
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1454

1455
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1456
		goto out_err;
1457

1458 1459
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1460

1461 1462
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1463
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1464 1465
	return;
out_err:
1466 1467
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1468
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1469 1470 1471 1472 1473 1474
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1475
/*
1476
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1477 1478 1479 1480
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1481 1482
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1483

1484
	dprintk("--> %s slot %p\n", __func__, slot);
1485

1486 1487
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1488

1489
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1490
	if (nfsd4_not_cached(resp)) {
1491
		slot->sl_datalen = 0;
1492
		return;
1493
	}
1494 1495 1496 1497 1498 1499 1500
	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;
1501 1502 1503
}

/*
1504 1505 1506 1507
 * 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.
1508 1509
 *
 */
1510 1511 1512
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1513
{
1514 1515
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1516

1517 1518 1519
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1520

1521
	/* Return nfserr_retry_uncached_rep in next operation. */
1522
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1523 1524 1525
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1526
	}
1527
	return op->status;
1528 1529 1530
}

/*
1531 1532
 * The sequence operation is not cached because we can use the slot and
 * session values.
1533 1534
 */
__be32
1535 1536
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1537
{
1538
	struct nfsd4_slot *slot = resp->cstate.slot;
1539 1540
	__be32 status;

1541
	dprintk("--> %s slot %p\n", __func__, slot);
1542

1543 1544 1545 1546
	/* 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;
1547

1548 1549
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1550

1551 1552 1553
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1554 1555 1556 1557

	return status;
}

A
Andy Adamson 已提交
1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573
/*
 * 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;
}

1574 1575 1576 1577 1578 1579 1580 1581
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.
	 */
1582 1583 1584
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1585 1586
}

A
Andy Adamson 已提交
1587 1588 1589 1590 1591
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1592
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1593
	__be32 status;
1594
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1595
	nfs4_verifier		verf = exid->verifier;
1596
	struct sockaddr		*sa = svc_addr(rqstp);
1597
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1598
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1599

1600
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1601
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1602
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1603
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1604
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1605

1606
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1607 1608 1609 1610 1611 1612
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1613 1614
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1615 1616 1617 1618 1619
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1620
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1621
	nfs4_lock_state();
1622
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1623
	if (conf) {
1624 1625 1626
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1627 1628
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1629
				status = nfserr_inval;
1630 1631
				goto out;
			}
1632
			if (!creds_match) { /* case 9 */
1633
				status = nfserr_perm;
1634 1635 1636
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1637 1638 1639
				status = nfserr_not_same;
				goto out;
			}
1640 1641 1642 1643
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1644
		}
1645
		if (!creds_match) { /* case 3 */
1646 1647
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1648 1649 1650 1651 1652
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1653
		if (verfs_match) { /* case 2 */
1654
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1655 1656 1657 1658 1659
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1660 1661
	}

1662
	if (update) { /* case 7 */
1663 1664
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1665 1666
	}

1667
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1668
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1669 1670
		expire_client(unconf);

1671
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1672
out_new:
J
Jeff Layton 已提交
1673
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1674
	if (new == NULL) {
1675
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1676 1677
		goto out;
	}
1678
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1679

1680
	gen_clid(new, nn);
1681
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1682 1683 1684 1685
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1686
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1687 1688 1689
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1690
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1691 1692 1693 1694 1695
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1696 1697
}

J
J. Bruce Fields 已提交
1698
static __be32
1699
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1700
{
1701 1702
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1703 1704

	/* The slot is in use, and no response has been sent. */
1705 1706
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1707 1708 1709 1710
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1711
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1712
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1713
		return nfs_ok;
1714
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1715 1716 1717 1718
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1719 1720 1721 1722 1723 1724 1725
/*
 * 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
1726
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739
{
	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;
}

1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756
#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
1757
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1758 1759 1760 1761 1762
{
	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
1763 1764 1765 1766 1767
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1768
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1769
	struct nfs4_client *conf, *unconf;
1770
	struct nfsd4_session *new;
1771
	struct nfsd4_conn *conn;
1772
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1773
	__be32 status = 0;
1774
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1775

1776 1777
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1778 1779
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1780
	new = alloc_session(&cr_ses->fore_channel, nn);
1781 1782 1783 1784 1785 1786
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1787

A
Andy Adamson 已提交
1788
	nfs4_lock_state();
1789
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1790
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
1791
	WARN_ON_ONCE(conf && unconf);
A
Andy Adamson 已提交
1792 1793

	if (conf) {
1794 1795
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1796
		if (status == nfserr_replay_cache) {
1797
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1798
			goto out_free_conn;
1799
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1800
			status = nfserr_seq_misordered;
1801
			goto out_free_conn;
A
Andy Adamson 已提交
1802 1803
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1804
		struct nfs4_client *old;
A
Andy Adamson 已提交
1805
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1806
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1807
			status = nfserr_clid_inuse;
1808
			goto out_free_conn;
A
Andy Adamson 已提交
1809
		}
1810 1811
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1812 1813
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1814
			status = nfserr_seq_misordered;
1815
			goto out_free_conn;
A
Andy Adamson 已提交
1816
		}
1817
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
J
J. Bruce Fields 已提交
1818 1819 1820
		if (old)
			expire_client(old);
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1821 1822 1823
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1824
		goto out_free_conn;
A
Andy Adamson 已提交
1825
	}
1826
	status = nfs_ok;
1827 1828 1829 1830 1831 1832
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1833 1834 1835
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1836
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1837
	       NFS4_MAX_SESSIONID_LEN);
1838 1839
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1840
	cs_slot->sl_seqid++;
1841
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1842

1843 1844
	/* 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
1845 1846
	nfs4_unlock_state();
	return status;
1847
out_free_conn:
1848
	nfs4_unlock_state();
1849 1850 1851
	free_conn(conn);
out_free_session:
	__free_session(new);
J
J. Bruce Fields 已提交
1852
	return status;
A
Andy Adamson 已提交
1853 1854
}

1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868
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;
}

1869 1870 1871
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1872
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1873

1874
	spin_lock(&nn->client_lock);
1875 1876
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1877
	spin_unlock(&nn->client_lock);
1878 1879 1880 1881 1882 1883

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1884 1885 1886 1887 1888
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1889
	struct nfsd4_conn *conn;
1890
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1891 1892 1893

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1894
	spin_lock(&nn->client_lock);
1895
	cstate->session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1896 1897 1898 1899 1900 1901
	/* Sorta weird: we only need the refcnt'ing because new_conn acquires
	 * client_lock iself: */
	if (cstate->session) {
		nfsd4_get_session(cstate->session);
		atomic_inc(&cstate->session->se_client->cl_refcount);
	}
1902
	spin_unlock(&nn->client_lock);
1903 1904 1905 1906
	if (!cstate->session)
		return nfserr_badsession;

	status = nfsd4_map_bcts_dir(&bcts->dir);
1907 1908 1909 1910 1911
	if (status)
		return status;
	conn = alloc_conn(rqstp, bcts->dir);
	if (!conn)
		return nfserr_jukebox;
1912
	nfsd4_init_conn(rqstp, conn, cstate->session);
1913
	return nfs_ok;
1914 1915
}

1916 1917 1918 1919 1920 1921 1922
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
1923 1924 1925 1926 1927
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1928
	struct nfsd4_session *ses;
J
J. Bruce Fields 已提交
1929
	__be32 status = nfserr_badsession;
1930
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1931 1932 1933 1934 1935 1936 1937 1938

	/* Notes:
	 * - The confirmed nfs4_client->cl_sessionid holds destroyed sessinid
	 * - Should we return nfserr_back_chan_busy if waiting for
	 *   callbacks on to-be-destroyed session?
	 * - Do we need to clear any callback info from previous session?
	 */

1939
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1940 1941 1942
		if (!nfsd4_last_compound_op(r))
			return nfserr_not_only_op;
	}
B
Benny Halevy 已提交
1943
	dump_sessionid(__func__, &sessionid->sessionid);
1944
	spin_lock(&nn->client_lock);
1945
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
B
Benny Halevy 已提交
1946
	if (!ses) {
1947
		spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1948 1949 1950 1951
		goto out;
	}

	unhash_session(ses);
1952
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1953

1954
	nfs4_lock_state();
1955
	nfsd4_probe_callback_sync(ses->se_client);
1956
	nfs4_unlock_state();
1957

1958
	spin_lock(&nn->client_lock);
1959
	nfsd4_del_conns(ses);
1960
	nfsd4_put_session_locked(ses);
1961
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1962 1963 1964
	status = nfs_ok;
out:
	return status;
A
Andy Adamson 已提交
1965 1966
}

1967
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
1968 1969 1970 1971
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
1972
		if (c->cn_xprt == xpt) {
1973 1974 1975 1976 1977 1978
			return c;
		}
	}
	return NULL;
}

1979
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
1980 1981
{
	struct nfs4_client *clp = ses->se_client;
1982
	struct nfsd4_conn *c;
1983
	int ret;
1984 1985

	spin_lock(&clp->cl_lock);
1986
	c = __nfsd4_find_conn(new->cn_xprt, ses);
1987 1988 1989 1990 1991 1992 1993
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
1994 1995 1996 1997
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
1998 1999 2000
	return;
}

2001 2002 2003 2004 2005 2006 2007
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
2008 2009 2010 2011 2012 2013 2014 2015
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
2016
__be32
B
Benny Halevy 已提交
2017
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2018 2019 2020
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2021
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2022 2023
	struct nfsd4_session *session;
	struct nfsd4_slot *slot;
2024
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2025
	__be32 status;
2026
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2027

2028 2029 2030
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2031 2032 2033 2034 2035 2036 2037 2038
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2039
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2040
	status = nfserr_badsession;
2041
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2042 2043 2044
	if (!session)
		goto out;

2045 2046 2047 2048
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
		goto out;

M
Mi Jinlong 已提交
2049 2050 2051 2052
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
		goto out;

B
Benny Halevy 已提交
2053
	status = nfserr_badslot;
2054
	if (seq->slotid >= session->se_fchannel.maxreqs)
B
Benny Halevy 已提交
2055 2056
		goto out;

2057
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2058 2059
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2060 2061 2062 2063 2064
	/* 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;

2065 2066
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2067
	if (status == nfserr_replay_cache) {
2068 2069 2070
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
			goto out;
B
Benny Halevy 已提交
2071 2072
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2073
		/* Return the cached reply status and set cstate->status
2074
		 * for nfsd4_proc_compound processing */
2075
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2076
		cstate->status = nfserr_replay_cache;
2077
		goto out;
B
Benny Halevy 已提交
2078 2079 2080 2081
	}
	if (status)
		goto out;

2082 2083
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2084

B
Benny Halevy 已提交
2085 2086
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2087
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2088 2089
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2090 2091
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2092 2093 2094 2095 2096

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

out:
2097
	/* Hold a session reference until done processing the compound. */
2098
	if (cstate->session) {
2099 2100
		struct nfs4_client *clp = session->se_client;

2101
		nfsd4_get_session(cstate->session);
2102
		atomic_inc(&clp->cl_refcount);
2103 2104
		switch (clp->cl_cb_state) {
		case NFSD4_CB_DOWN:
2105
			seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
2106 2107
			break;
		case NFSD4_CB_FAULT:
2108
			seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
2109
			break;
2110 2111
		default:
			seq->status_flags = 0;
2112
		}
2113
	}
2114
	kfree(conn);
2115
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2116
	return status;
A
Andy Adamson 已提交
2117 2118
}

2119 2120 2121 2122
__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
2123
	__be32 status = 0;
2124
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2125 2126

	nfs4_lock_state();
2127
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2128
	conf = find_confirmed_client(&dc->clientid, true, nn);
2129
	WARN_ON_ONCE(conf && unconf);
2130 2131 2132 2133

	if (conf) {
		clp = conf;

2134
		if (client_has_state(conf)) {
2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150
			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;
}

2151 2152 2153
__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
2154
	__be32 status = 0;
2155

2156 2157 2158 2159 2160 2161 2162 2163 2164
	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;
	}
2165

2166
	nfs4_lock_state();
2167
	status = nfserr_complete_already;
2168 2169
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2170 2171 2172 2173
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2174 2175 2176 2177 2178 2179 2180
		/*
		 * 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.
		 */
2181 2182 2183
		goto out;

	status = nfs_ok;
2184
	nfsd4_client_record_create(cstate->session->se_client);
2185
out:
2186
	nfs4_unlock_state();
2187
	return status;
2188 2189
}

2190
__be32
2191 2192
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2193
{
2194
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2195
	nfs4_verifier		clverifier = setclid->se_verf;
2196
	struct nfs4_client	*conf, *unconf, *new;
2197
	__be32 			status;
2198 2199
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2200
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2201
	nfs4_lock_state();
2202
	conf = find_confirmed_client_by_name(&clname, nn);
2203
	if (conf) {
2204
		/* case 0: */
L
Linus Torvalds 已提交
2205
		status = nfserr_clid_inuse;
2206 2207
		if (clp_used_exchangeid(conf))
			goto out;
2208
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2209 2210 2211 2212 2213
			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
2214 2215 2216
			goto out;
		}
	}
2217
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2218 2219
	if (unconf)
		expire_client(unconf);
2220
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2221
	new = create_client(clname, rqstp, &clverifier);
2222 2223
	if (new == NULL)
		goto out;
2224
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2225
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2226
		copy_clid(new, conf);
2227
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2228
		gen_clid(new, nn);
2229
	new->cl_minorversion = 0;
2230
	gen_callback(new, setclid, rqstp);
2231
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2232 2233 2234 2235 2236 2237 2238 2239 2240 2241
	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;
}


2242
__be32
2243 2244 2245
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2246
{
2247
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2248 2249
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2250
	__be32 status;
2251
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2252

2253
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2254 2255
		return nfserr_stale_clientid;
	nfs4_lock_state();
2256

2257
	conf = find_confirmed_client(clid, false, nn);
2258
	unconf = find_unconfirmed_client(clid, false, nn);
2259
	/*
2260 2261 2262 2263
	 * 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.
2264
	 */
2265 2266 2267 2268 2269
	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;
2270
	/* cases below refer to rfc 3530 section 14.2.34: */
2271 2272
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2273
			status = nfs_ok;
2274 2275 2276 2277 2278 2279
		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 */
2280 2281 2282
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2283
	} else { /* case 3: normal case; new or rebooted client */
2284
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2285
		if (conf)
2286 2287
			expire_client(conf);
		move_to_confirmed(unconf);
2288
		nfsd4_probe_callback(unconf);
2289
	}
L
Linus Torvalds 已提交
2290 2291 2292 2293 2294
out:
	nfs4_unlock_state();
	return status;
}

2295 2296 2297 2298 2299
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2300
/* OPEN Share state helper functions */
2301
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2302 2303 2304
{
	unsigned int hashval = file_hashval(ino);

2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316
	atomic_set(&fp->fi_ref, 1);
	INIT_LIST_HEAD(&fp->fi_hash);
	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);
	list_add(&fp->fi_hash, &file_hashtbl[hashval]);
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2317 2318
}

N
NeilBrown 已提交
2319
static void
2320
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2321
{
N
NeilBrown 已提交
2322 2323
	if (*slab == NULL)
		return;
2324
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2325
	*slab = NULL;
L
Linus Torvalds 已提交
2326 2327
}

2328
void
L
Linus Torvalds 已提交
2329 2330
nfsd4_free_slabs(void)
{
2331 2332
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2333
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2334
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2335
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2336
}
L
Linus Torvalds 已提交
2337

2338
int
N
NeilBrown 已提交
2339 2340
nfsd4_init_slabs(void)
{
2341 2342 2343 2344 2345
	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",
2346
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2347
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2348 2349
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2350
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2351 2352
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2353
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2354
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2355 2356
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2357
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2358
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2359 2360
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2361 2362 2363 2364 2365
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2366 2367
}

2368 2369 2370 2371 2372 2373 2374
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
2375
{
2376 2377
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2378 2379
}

2380
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2381
{
2382 2383 2384
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2385 2386
}

2387
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2388
{
L
Linus Torvalds 已提交
2389 2390
	struct nfs4_stateowner *sop;

2391
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2392 2393 2394 2395 2396
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2397
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2398
		return NULL;
2399 2400 2401
	}
	sop->so_owner.len = owner->len;

2402
	INIT_LIST_HEAD(&sop->so_stateids);
2403 2404 2405 2406 2407
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2408
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2409
{
2410 2411 2412
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2413
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2414 2415
}

2416
static struct nfs4_openowner *
2417
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2418
	struct nfs4_openowner *oo;
2419

2420 2421
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2422
		return NULL;
2423 2424
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2425
	oo->oo_flags = NFS4_OO_NEW;
2426
	oo->oo_time = 0;
2427
	oo->oo_last_closed_stid = NULL;
2428 2429 2430
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2431 2432
}

2433
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2434
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2435

J
J. Bruce Fields 已提交
2436
	stp->st_stid.sc_type = NFS4_OPEN_STID;
2437
	INIT_LIST_HEAD(&stp->st_lockowners);
2438
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2439
	list_add(&stp->st_perfile, &fp->fi_stateids);
2440
	stp->st_stateowner = &oo->oo_owner;
2441
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2442 2443 2444
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2445
	set_access(open->op_share_access, stp);
2446
	set_deny(open->op_share_deny, stp);
2447
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2448 2449
}

2450
static void
2451
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2452
{
2453 2454
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2455
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2456

2457
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2458
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2459 2460 2461
}

static int
2462 2463 2464 2465 2466 2467
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
2468 2469
}

2470
static struct nfs4_openowner *
2471 2472
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2473
{
2474 2475
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2476
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2477

2478
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2479 2480
		if (!so->so_is_open_owner)
			continue;
2481 2482
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2483 2484 2485
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2486 2487 2488
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499
	}
	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;

2500
	spin_lock(&recall_lock);
L
Linus Torvalds 已提交
2501
	list_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2502 2503
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2504
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2505
			return fp;
2506
		}
L
Linus Torvalds 已提交
2507
	}
2508
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2509 2510 2511 2512 2513 2514 2515
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2516
static __be32
L
Linus Torvalds 已提交
2517 2518 2519 2520
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;
2521
	struct nfs4_ol_stateid *stp;
2522
	__be32 ret;
L
Linus Torvalds 已提交
2523 2524

	fp = find_file(ino);
2525 2526
	if (!fp)
		return nfs_ok;
2527
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2528
	/* Search for conflicting share reservations */
2529
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2530 2531
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2532
			goto out;
L
Linus Torvalds 已提交
2533
	}
2534 2535 2536 2537
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2538 2539
}

2540
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2541 2542 2543 2544 2545 2546 2547 2548 2549 2550
{
	/* 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);

2551
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2552 2553
	dp->dl_time = get_seconds();

2554 2555 2556
	nfsd4_cb_recall(dp);
}

2557
/* Called from break_lease() with lock_flocks() held. */
2558 2559
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2560 2561
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2562

2563 2564 2565 2566 2567 2568 2569 2570
	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;
	}
2571 2572
	/*
	 * We don't want the locks code to timeout the lease for us;
2573
	 * we'll remove it ourself if a delegation isn't returned
2574
	 * in time:
2575 2576
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2577

2578
	spin_lock(&recall_lock);
2579 2580 2581
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2582
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593
}

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

2594
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2595 2596
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2597 2598
};

2599 2600 2601 2602 2603 2604 2605 2606 2607 2608
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
2609

2610
__be32
A
Andy Adamson 已提交
2611
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2612
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2613 2614 2615 2616
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2617
	struct nfs4_openowner *oo = NULL;
2618
	__be32 status;
L
Linus Torvalds 已提交
2619

2620
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2621
		return nfserr_stale_clientid;
2622 2623 2624 2625 2626 2627 2628
	/*
	 * 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
2629

2630
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2631
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2632 2633
	open->op_openowner = oo;
	if (!oo) {
2634 2635
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2636
		if (clp == NULL)
2637
			return nfserr_expired;
2638
		goto new_owner;
L
Linus Torvalds 已提交
2639
	}
2640
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2641
		/* Replace unconfirmed owners without checking for replay. */
2642 2643 2644
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2645
		goto new_owner;
2646
	}
2647 2648 2649 2650 2651
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2652 2653 2654 2655 2656
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2657 2658 2659 2660
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2661
	return nfs_ok;
L
Linus Torvalds 已提交
2662 2663
}

2664
static inline __be32
N
NeilBrown 已提交
2665 2666 2667 2668 2669 2670 2671 2672
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;
}

2673
static int share_access_to_flags(u32 share_access)
2674
{
2675
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2676 2677
}

2678
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2679
{
2680
	struct nfs4_stid *ret;
2681

2682
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2683 2684 2685
	if (!ret)
		return NULL;
	return delegstateid(ret);
2686 2687
}

2688 2689 2690 2691 2692 2693
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;
}

2694
static __be32
2695
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2696 2697 2698
		struct nfs4_delegation **dp)
{
	int flags;
2699
	__be32 status = nfserr_bad_stateid;
2700

2701
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2702
	if (*dp == NULL)
2703
		goto out;
2704
	flags = share_access_to_flags(open->op_share_access);
2705 2706 2707
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2708
out:
2709
	if (!nfsd4_is_deleg_cur(open))
2710 2711 2712
		return nfs_ok;
	if (status)
		return status;
2713
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2714
	return nfs_ok;
2715 2716
}

2717
static __be32
2718
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2719
{
2720
	struct nfs4_ol_stateid *local;
2721
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2722

2723
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2724 2725 2726 2727
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2728
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2729 2730 2731
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2732
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2733
	}
2734
	return nfs_ok;
L
Linus Torvalds 已提交
2735 2736
}

2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747
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;
}

2748 2749
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2750 2751
{
	__be32 status;
2752 2753 2754
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765
	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;
}

2766
static inline __be32
L
Linus Torvalds 已提交
2767 2768 2769 2770 2771 2772 2773 2774 2775 2776
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))
2777
		return nfserr_inval;
L
Linus Torvalds 已提交
2778 2779 2780
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2781
static __be32
2782
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
2783
{
2784
	u32 op_share_access = open->op_share_access;
2785
	bool new_access;
2786
	__be32 status;
L
Linus Torvalds 已提交
2787

2788
	new_access = !test_access(op_share_access, stp);
2789
	if (new_access) {
2790
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2791 2792
		if (status)
			return status;
2793
	}
L
Linus Torvalds 已提交
2794 2795
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2796
		if (new_access) {
2797
			int oflag = nfs4_access_to_omode(op_share_access);
2798 2799
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2800 2801 2802
		return status;
	}
	/* remember the open */
2803
	set_access(op_share_access, stp);
2804
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2805 2806 2807 2808 2809 2810

	return nfs_ok;
}


static void
2811
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2812
{
2813
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2814 2815
}

2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828
/* 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;
}

2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840
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;
2841
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2842 2843 2844 2845
	fl->fl_pid = current->tgid;
	return fl;
}

2846 2847
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2848
	struct nfs4_file *fp = dp->dl_file;
2849 2850 2851 2852 2853 2854
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2855
	fl->fl_file = find_readable_file(fp);
2856
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2857
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2858
	if (status) {
2859
		list_del_init(&dp->dl_perclnt);
2860 2861 2862
		locks_free_lock(fl);
		return -ENOMEM;
	}
2863
	fp->fi_lease = fl;
A
Al Viro 已提交
2864
	fp->fi_deleg_file = get_file(fl->fl_file);
2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
	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);
2884
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2885 2886 2887
	return 0;
}

2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903
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:
2904
			WARN_ON_ONCE(1);
2905 2906 2907 2908
		}
	}
}

L
Linus Torvalds 已提交
2909 2910 2911 2912
/*
 * Attempt to hand out a delegation.
 */
static void
2913 2914
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
2915 2916
{
	struct nfs4_delegation *dp;
2917
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2918
	int cb_up;
2919
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2920

2921
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2922
	flag = NFS4_OPEN_DELEGATE_NONE;
2923 2924 2925
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2926
			if (!cb_up)
2927 2928 2929 2930 2931 2932 2933 2934
				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.... */
2935
			if (locks_in_grace(net))
2936
				goto out;
2937
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2938 2939 2940 2941 2942 2943 2944 2945 2946
				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
2947

2948
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2949 2950
	if (dp == NULL)
		goto out_no_deleg;
2951
	status = nfs4_set_delegation(dp, flag);
2952
	if (status)
2953
		goto out_free;
L
Linus Torvalds 已提交
2954

2955
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2956

2957
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
2958
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2959 2960
out:
	open->op_delegate_type = flag;
2961 2962 2963 2964 2965
	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");

2966 2967 2968
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
2969
	}
2970 2971
	return;
out_free:
2972
	unhash_stid(&dp->dl_stid);
2973
	nfs4_put_delegation(dp);
2974 2975 2976
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
2977 2978
}

2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996
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
2997 2998 2999
/*
 * called with nfs4_lock_state() held.
 */
3000
__be32
L
Linus Torvalds 已提交
3001 3002
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3003
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3004
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3005 3006
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3007
	struct nfs4_ol_stateid *stp = NULL;
3008
	struct nfs4_delegation *dp = NULL;
3009
	__be32 status;
L
Linus Torvalds 已提交
3010 3011 3012 3013 3014 3015 3016 3017 3018 3019

	/*
	 * 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;
3020
		status = nfs4_check_deleg(cl, fp, open, &dp);
3021 3022
		if (status)
			goto out;
L
Linus Torvalds 已提交
3023
	} else {
3024
		status = nfserr_bad_stateid;
3025
		if (nfsd4_is_deleg_cur(open))
3026
			goto out;
3027
		status = nfserr_jukebox;
3028 3029 3030
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3031 3032 3033 3034 3035 3036 3037 3038
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3039
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3040 3041 3042
		if (status)
			goto out;
	} else {
3043
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3044 3045 3046
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3047
		if (status)
L
Linus Torvalds 已提交
3048
			goto out;
3049 3050
		stp = open->op_stp;
		open->op_stp = NULL;
3051
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3052
	}
3053 3054
	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
3055

3056
	if (nfsd4_has_session(&resp->cstate)) {
3057
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3058

3059 3060 3061 3062 3063 3064 3065
		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
3066 3067 3068 3069
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3070
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3071
nodeleg:
L
Linus Torvalds 已提交
3072 3073
	status = nfs_ok;

3074
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3075
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3076
out:
3077 3078
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3079 3080
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3081

3082 3083
	if (fp)
		put_nfs4_file(fp);
3084
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3085
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3086 3087 3088 3089
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3090
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3091
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3092 3093 3094 3095 3096
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111
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;
		}
	}
3112 3113
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3114
	if (open->op_stp)
3115
		free_generic_stateid(open->op_stp);
3116 3117
}

3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129
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;
}

3130
__be32
3131 3132
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3133 3134
{
	struct nfs4_client *clp;
3135
	__be32 status;
3136
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3137 3138 3139 3140

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3141 3142
	status = lookup_clientid(clid, cstate->minorversion, nn, &clp);
	if (status)
L
Linus Torvalds 已提交
3143 3144
		goto out;
	status = nfserr_cb_path_down;
3145
	if (!list_empty(&clp->cl_delegations)
3146
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3147 3148 3149 3150 3151 3152 3153
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3154
static void
3155
nfsd4_end_grace(struct nfsd_net *nn)
3156
{
3157
	/* do nothing if grace period already ended */
3158
	if (nn->grace_ended)
3159 3160
		return;

3161
	dprintk("NFSD: end of grace period\n");
3162
	nn->grace_ended = true;
3163
	nfsd4_record_grace_done(nn, nn->boot_time);
3164
	locks_end_grace(&nn->nfsd4_manager);
3165 3166 3167 3168 3169
	/*
	 * 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:
	 */
3170
	nn->nfsd4_grace = nn->nfsd4_lease;
3171 3172
}

3173
static time_t
3174
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3175 3176
{
	struct nfs4_client *clp;
3177
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3178 3179
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3180 3181 3182
	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
3183 3184 3185 3186

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3187
	nfsd4_end_grace(nn);
3188
	INIT_LIST_HEAD(&reaplist);
3189
	spin_lock(&nn->client_lock);
3190
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3191 3192 3193 3194 3195 3196 3197
		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;
		}
3198 3199 3200 3201 3202 3203 3204
		if (atomic_read(&clp->cl_refcount)) {
			dprintk("NFSD: client in use (clientid %08x)\n",
				clp->cl_clientid.cl_id);
			continue;
		}
		unhash_client_locked(clp);
		list_add(&clp->cl_lru, &reaplist);
3205
	}
3206
	spin_unlock(&nn->client_lock);
3207 3208
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3209 3210 3211 3212 3213 3214 3215
		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);
3216 3217
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230
		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);
	}
3231
	test_val = nn->nfsd4_lease;
3232
	list_for_each_safe(pos, next, &nn->close_lru) {
3233 3234 3235
		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
3236 3237 3238 3239
			if (test_val > u)
				test_val = u;
			break;
		}
3240
		release_openowner(oo);
L
Linus Torvalds 已提交
3241 3242 3243 3244 3245 3246 3247
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3248 3249 3250 3251
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3252
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3253 3254
{
	time_t t;
3255 3256 3257 3258
	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
3259

3260
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3261
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3262
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3263 3264
}

3265
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3266
{
3267 3268 3269
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3270 3271 3272
}

static inline int
3273
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3274
{
3275 3276 3277
	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
3278 3279 3280
}

static inline int
3281
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3282
{
3283 3284
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3285 3286 3287
}

static
3288
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3289
{
3290
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3291

3292 3293 3294
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3295
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3296
                goto out;
3297
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3298 3299 3300 3301 3302 3303
                goto out;
	status = nfs_ok;
out:
	return status;
}

3304
static inline __be32
3305
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3306
{
3307
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3308
		return nfs_ok;
3309
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3310
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325
		 * 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
3326
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3327
{
3328
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3329 3330
}

3331 3332 3333 3334 3335 3336
/* 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
3337
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3338
{
A
Andy Adamson 已提交
3339 3340 3341 3342
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3343
	if (has_session && in->si_generation == 0)
3344 3345 3346 3347
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3348

3349
	/* If the client sends us a stateid from the future, it's buggy: */
3350
	if (stateid_generation_after(in, ref))
3351 3352
		return nfserr_bad_stateid;
	/*
3353 3354 3355 3356 3357 3358 3359 3360
	 * 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:
3361
	 */
3362
	return nfserr_old_stateid;
3363 3364
}

3365
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3366
{
3367 3368 3369
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3370

3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381
	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;
	}
3382
	s = find_stateid(cl, stateid);
3383
	if (!s)
3384
		return nfserr_bad_stateid;
3385
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3386
	if (status)
3387 3388 3389 3390 3391
		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
3392
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3393 3394
		return nfserr_bad_stateid;
	return nfs_ok;
3395 3396
}

3397 3398 3399
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3400 3401
{
	struct nfs4_client *cl;
J
J. Bruce Fields 已提交
3402
	__be32 status;
3403 3404 3405

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
J
J. Bruce Fields 已提交
3406 3407 3408
	status = lookup_clientid(&stateid->si_opaque.so_clid, sessions,
							nn, &cl);
	if (status == nfserr_stale_clientid)
3409
		return nfserr_stale_stateid;
J
J. Bruce Fields 已提交
3410 3411
	if (status)
		return status;
3412 3413 3414 3415 3416 3417
	*s = find_stateid_by_type(cl, stateid, typemask);
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;
}

L
Linus Torvalds 已提交
3418 3419 3420
/*
* Checks for stateid operations
*/
3421
__be32
3422
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3423
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3424
{
3425
	struct nfs4_stid *s;
3426
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3427
	struct nfs4_delegation *dp = NULL;
3428
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3429
	struct inode *ino = current_fh->fh_dentry->d_inode;
3430
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3431
	__be32 status;
L
Linus Torvalds 已提交
3432 3433 3434 3435

	if (filpp)
		*filpp = NULL;

3436
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3437 3438 3439
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3440
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3441

3442 3443
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3444 3445
	if (status)
		return status;
3446 3447 3448
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3449 3450
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3451
		dp = delegstateid(s);
3452 3453 3454
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3455
		if (filpp) {
3456
			*filpp = dp->dl_file->fi_deleg_file;
3457 3458 3459 3460 3461
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3462
		}
3463 3464 3465
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3466
		stp = openlockstateid(s);
3467 3468
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3469
			goto out;
3470
		if (stp->st_stateowner->so_is_open_owner
3471
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3472
			goto out;
3473 3474
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3475
			goto out;
3476 3477 3478 3479 3480 3481
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3482 3483 3484
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3485 3486 3487 3488 3489 3490
	}
	status = nfs_ok;
out:
	return status;
}

3491
static __be32
3492
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3493
{
3494
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3495 3496 3497 3498 3499
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3500 3501 3502 3503 3504 3505 3506
/*
 * 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)
{
3507 3508 3509 3510 3511
	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)
3512 3513
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3514 3515
	nfs4_unlock_state();

3516 3517 3518
	return nfs_ok;
}

3519 3520 3521 3522 3523
__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
3524
	struct nfs4_stid *s;
3525
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3526
	__be32 ret = nfserr_bad_stateid;
3527 3528

	nfs4_lock_state();
3529
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3530
	if (!s)
3531
		goto out;
J
J. Bruce Fields 已提交
3532 3533
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3534 3535
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3536 3537 3538 3539 3540 3541 3542 3543 3544
	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;
3545 3546 3547
		break;
	default:
		ret = nfserr_bad_stateid;
3548 3549 3550 3551 3552 3553
	}
out:
	nfs4_unlock_state();
	return ret;
}

3554 3555 3556 3557 3558 3559
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
3560

3561
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3562 3563 3564 3565 3566 3567 3568 3569
{
	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;
3570 3571 3572 3573 3574 3575 3576 3577 3578 3579
	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);
3580 3581
}

L
Linus Torvalds 已提交
3582 3583 3584
/* 
 * Checks for sequence id mutating operations. 
 */
3585
static __be32
3586
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3587
			 stateid_t *stateid, char typemask,
3588 3589
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3590
{
3591
	__be32 status;
3592
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3593

3594 3595
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3596

L
Linus Torvalds 已提交
3597
	*stpp = NULL;
3598 3599
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3600 3601
	if (status)
		return status;
3602
	*stpp = openlockstateid(s);
3603
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3604

3605 3606
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3607

3608 3609
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)
3610 3611 3612
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3613

3614
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3615
						NFS4_OPEN_STID, stpp, nn);
3616 3617
	if (status)
		return status;
3618
	oo = openowner((*stpp)->st_stateowner);
3619
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3620 3621
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3622 3623
}

3624
__be32
3625
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3626
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3627
{
3628
	__be32 status;
3629
	struct nfs4_openowner *oo;
3630
	struct nfs4_ol_stateid *stp;
3631
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3632 3633

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3634 3635
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3636

3637
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3638 3639
	if (status)
		return status;
L
Linus Torvalds 已提交
3640 3641 3642

	nfs4_lock_state();

3643
	status = nfs4_preprocess_seqid_op(cstate,
3644
					oc->oc_seqid, &oc->oc_req_stateid,
3645
					NFS4_OPEN_STID, &stp, nn);
3646
	if (status)
3647
		goto out;
3648
	oo = openowner(stp->st_stateowner);
3649
	status = nfserr_bad_stateid;
3650
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3651
		goto out;
3652
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3653 3654
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3655
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3656
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3657

3658
	nfsd4_client_record_create(oo->oo_owner.so_client);
3659
	status = nfs_ok;
L
Linus Torvalds 已提交
3660
out:
3661 3662
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3663 3664 3665
	return status;
}

J
J. Bruce Fields 已提交
3666
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3667
{
3668
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3669 3670
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3671
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3672
}
3673

J
J. Bruce Fields 已提交
3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687
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:
3688
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3689 3690 3691 3692
	}
}

static void
3693
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3694 3695 3696 3697
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3698
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3699 3700 3701
	}
}

3702
__be32
3703 3704
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3705
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3706
{
3707
	__be32 status;
3708
	struct nfs4_ol_stateid *stp;
3709
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3710 3711

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3712 3713
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3714

3715
	/* We don't yet support WANT bits: */
3716 3717 3718
	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
3719 3720

	nfs4_lock_state();
3721
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3722
					&od->od_stateid, &stp, nn);
3723
	if (status)
L
Linus Torvalds 已提交
3724 3725
		goto out; 
	status = nfserr_inval;
3726 3727
	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
3728 3729 3730
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3731
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3732 3733 3734 3735
		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
3736
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3737

3738
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3739

3740 3741
	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
3742 3743
	status = nfs_ok;
out:
3744 3745
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3746 3747 3748
	return status;
}

3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765
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;
3766 3767 3768 3769 3770 3771 3772
	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;
3773 3774
}

L
Linus Torvalds 已提交
3775 3776 3777
/*
 * nfs4_unlock_state() called after encode
 */
3778
__be32
3779
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3780
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3781
{
3782
	__be32 status;
3783
	struct nfs4_openowner *oo;
3784
	struct nfs4_ol_stateid *stp;
3785 3786
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3787 3788

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3789 3790
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3791 3792

	nfs4_lock_state();
3793 3794 3795
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3796
					&stp, nn);
3797
	if (status)
L
Linus Torvalds 已提交
3798
		goto out; 
3799
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3800
	status = nfs_ok;
3801 3802
	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
3803

3804
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3805
	release_last_closed_stateid(oo);
3806
	oo->oo_flags &= ~NFS4_OO_PURGE_CLOSE;
3807
	oo->oo_last_closed_stid = stp;
3808

3809 3810 3811 3812 3813 3814 3815 3816 3817 3818
	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))
3819
				move_to_close_lru(oo, SVC_NET(rqstp));
3820 3821
		}
	}
L
Linus Torvalds 已提交
3822
out:
3823 3824
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3825 3826 3827
	return status;
}

3828
__be32
3829 3830
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3831
{
3832 3833
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3834
	struct nfs4_stid *s;
3835
	__be32 status;
3836
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3837

3838
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3839
		return status;
L
Linus Torvalds 已提交
3840 3841

	nfs4_lock_state();
3842 3843
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3844
	if (status)
3845
		goto out;
3846
	dp = delegstateid(s);
3847
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3848 3849 3850 3851
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3852
out:
3853 3854
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3855 3856 3857 3858 3859 3860
	return status;
}


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

3861
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3862

B
Benny Halevy 已提交
3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877
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;

3878
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3879 3880 3881 3882
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3883
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3884 3885 3886
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3887
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906
}

/*
 * 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;
}

3907 3908
/* 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. */
3909
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3910
};
L
Linus Torvalds 已提交
3911 3912 3913 3914

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3915
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3916

3917
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3918 3919 3920
		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);
3921 3922 3923
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3924 3925
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3926
	} else {
3927 3928 3929
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3930 3931
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3932 3933
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3934 3935
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3936 3937 3938 3939 3940 3941
		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
3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952
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;
}

3953 3954
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3955
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3956
{
3957
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
3958
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3959

3960
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
3961 3962
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
3963 3964 3965 3966
	}
	return NULL;
}

3967
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
3968
{
3969 3970 3971
	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);
3972
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
3973

3974
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
3975
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
3976
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
3977 3978
}

L
Linus Torvalds 已提交
3979 3980 3981
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
3982
 * occurred. 
L
Linus Torvalds 已提交
3983
 *
3984
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
3985 3986
 */

3987
static struct nfs4_lockowner *
3988
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
3989
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3990

3991 3992
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
3993
		return NULL;
3994 3995
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
3996 3997
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
3998 3999 4000
	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
4001 4002
}

4003 4004
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
4005
{
4006
	struct nfs4_ol_stateid *stp;
4007
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4008

4009
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4010
	if (stp == NULL)
J
J. Bruce Fields 已提交
4011
		return NULL;
J
J. Bruce Fields 已提交
4012
	stp->st_stid.sc_type = NFS4_LOCK_STID;
4013
	list_add(&stp->st_perfile, &fp->fi_stateids);
4014 4015
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4016
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4017
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4018
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4019
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4020
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4021 4022 4023
	return stp;
}

4024
static int
L
Linus Torvalds 已提交
4025 4026
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4027
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4028 4029 4030
	     LOFF_OVERFLOW(offset, length)));
}

4031
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4032 4033 4034 4035
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4036
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4037 4038
		return;
	nfs4_file_get_access(fp, oflag);
4039
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4040 4041
}

J
J. Bruce Fields 已提交
4042
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)
4043 4044 4045 4046 4047 4048
{
	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;
4049
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4050

4051 4052
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4053 4054 4055 4056 4057 4058 4059 4060
	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;
	}
4061
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074
			&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
4075 4076 4077
/*
 *  LOCK operation 
 */
4078
__be32
4079
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4080
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4081
{
4082 4083
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4084
	struct nfs4_ol_stateid *lock_stp;
4085
	struct file *filp = NULL;
4086 4087
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4088
	__be32 status = 0;
4089
	bool new_state = false;
4090
	int lkflg;
4091
	int err;
4092 4093
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4094 4095 4096 4097 4098 4099 4100 4101

	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;

4102
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4103
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4104 4105 4106 4107
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4108 4109 4110
	nfs4_lock_state();

	if (lock->lk_is_new) {
4111
		struct nfs4_ol_stateid *open_stp = NULL;
4112 4113 4114 4115 4116 4117 4118

		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
4119
		status = nfserr_stale_clientid;
4120
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4121 4122 4123
			goto out;

		/* validate and update open stateid and open seqid */
4124
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4125 4126
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4127
					&open_stp, nn);
4128
		if (status)
L
Linus Torvalds 已提交
4129
			goto out;
4130
		open_sop = openowner(open_stp->st_stateowner);
4131
		status = nfserr_bad_stateid;
4132
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4133 4134
						&lock->v.new.clientid))
			goto out;
4135 4136
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4137
	} else
4138
		status = nfs4_preprocess_seqid_op(cstate,
4139 4140
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4141
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4142 4143
	if (status)
		goto out;
4144
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4145

4146 4147 4148 4149 4150
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4151
	status = nfserr_grace;
4152
	if (locks_in_grace(net) && !lock->lk_reclaim)
4153 4154
		goto out;
	status = nfserr_no_grace;
4155
	if (!locks_in_grace(net) && lock->lk_reclaim)
4156 4157
		goto out;

4158 4159 4160 4161 4162 4163 4164 4165
	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
4166 4167 4168
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4169 4170 4171
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4172
			file_lock->fl_type = F_RDLCK;
4173
			break;
L
Linus Torvalds 已提交
4174 4175
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4176 4177 4178
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4179
			file_lock->fl_type = F_WRLCK;
4180
			break;
L
Linus Torvalds 已提交
4181 4182 4183 4184
		default:
			status = nfserr_inval;
		goto out;
	}
4185 4186 4187 4188
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203
	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
4204

4205
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4206
	switch (-err) {
L
Linus Torvalds 已提交
4207
	case 0: /* success! */
4208 4209
		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
4210
				sizeof(stateid_t));
4211
		status = 0;
4212 4213 4214 4215
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4216
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4217
		break;
L
Linus Torvalds 已提交
4218 4219
	case (EDEADLK):
		status = nfserr_deadlock;
4220
		break;
4221
	default:
4222
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4223
		status = nfserrno(err);
4224
		break;
L
Linus Torvalds 已提交
4225 4226
	}
out:
4227
	if (status && new_state)
4228
		release_lockowner(lock_sop);
4229 4230
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4231 4232 4233 4234
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4235 4236 4237
	return status;
}

4238 4239 4240 4241 4242 4243
/*
 * 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.)
 */
4244
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4245 4246
{
	struct file *file;
4247 4248 4249 4250 4251
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4252 4253 4254
	return err;
}

L
Linus Torvalds 已提交
4255 4256 4257
/*
 * LOCKT operation
 */
4258
__be32
4259 4260
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4261 4262
{
	struct inode *inode;
4263
	struct file_lock *file_lock = NULL;
4264
	struct nfs4_lockowner *lo;
4265
	__be32 status;
4266
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4267

4268
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4269 4270 4271 4272 4273 4274 4275
		return nfserr_grace;

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

	nfs4_lock_state();

4276 4277 4278 4279 4280
	if (!nfsd4_has_session(cstate)) {
		status = lookup_clientid(&lockt->lt_clientid, false, nn, NULL);
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
4281

4282
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4283 4284
		goto out;

4285
	inode = cstate->current_fh.fh_dentry->d_inode;
4286 4287 4288 4289 4290 4291 4292
	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
4293 4294 4295
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4296
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4297 4298 4299
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4300
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4301 4302
		break;
		default:
4303
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4304 4305 4306 4307
			status = nfserr_inval;
		goto out;
	}

4308
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4309
	if (lo)
4310 4311 4312
		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
4313

4314 4315
	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
4316

4317
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4318

4319
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4320
	if (status)
4321
		goto out;
4322

4323
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4324
		status = nfserr_denied;
4325
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4326 4327 4328
	}
out:
	nfs4_unlock_state();
4329 4330
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4331 4332 4333
	return status;
}

4334
__be32
4335
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4336
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4337
{
4338
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4339
	struct file *filp = NULL;
4340
	struct file_lock *file_lock = NULL;
4341
	__be32 status;
4342
	int err;
4343 4344
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4345 4346 4347 4348 4349 4350 4351 4352 4353
	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();
									        
4354
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4355 4356
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4357
	if (status)
L
Linus Torvalds 已提交
4358
		goto out;
4359 4360 4361 4362 4363
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381
	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
4382 4383 4384 4385

	/*
	*  Try to unlock the file in the VFS.
	*/
4386
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4387
	if (err) {
4388
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4389 4390 4391 4392 4393
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4394 4395
	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
4396 4397

out:
4398 4399
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4400 4401
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4402 4403 4404
	return status;

out_nfserr:
4405
	status = nfserrno(err);
L
Linus Torvalds 已提交
4406 4407 4408 4409 4410 4411 4412 4413 4414
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4415
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4416 4417
{
	struct file_lock **flpp;
4418
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4419 4420
	int status = 0;

4421
	lock_flocks();
L
Linus Torvalds 已提交
4422
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4423
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4424 4425
			status = 1;
			goto out;
4426
		}
L
Linus Torvalds 已提交
4427 4428
	}
out:
4429
	unlock_flocks();
L
Linus Torvalds 已提交
4430 4431 4432
	return status;
}

4433
__be32
4434 4435 4436
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4437 4438
{
	clientid_t *clid = &rlockowner->rl_clientid;
4439
	struct nfs4_stateowner *sop;
4440
	struct nfs4_lockowner *lo;
4441
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4442
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4443
	struct list_head matches;
4444
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4445
	__be32 status;
4446
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4447 4448 4449 4450 4451 4452

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

	nfs4_lock_state();

4453 4454 4455 4456
	status = lookup_clientid(clid, cstate->minorversion, nn, NULL);
	if (status)
		goto out;

4457 4458
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4459

4460
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4461 4462
		if (sop->so_is_open_owner)
			continue;
4463 4464 4465 4466 4467 4468 4469 4470
		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
4471
		}
4472 4473 4474 4475 4476
	}
	/* 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
4477
	while (!list_empty(&matches)) {
4478 4479
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4480 4481
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4482 4483
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4484 4485 4486 4487 4488 4489 4490
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4491
alloc_reclaim(void)
L
Linus Torvalds 已提交
4492
{
N
NeilBrown 已提交
4493
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4494 4495
}

4496
bool
4497
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4498
{
4499
	struct nfs4_client_reclaim *crp;
4500

4501
	crp = nfsd4_find_reclaim_client(name, nn);
4502
	return (crp && crp->cr_clp);
4503 4504
}

L
Linus Torvalds 已提交
4505 4506 4507
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4508
struct nfs4_client_reclaim *
4509
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4510 4511
{
	unsigned int strhashval;
4512
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4513

N
NeilBrown 已提交
4514 4515
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4516 4517 4518
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4519
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4520
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4521
		crp->cr_clp = NULL;
4522
		nn->reclaim_str_hashtbl_size++;
4523 4524
	}
	return crp;
L
Linus Torvalds 已提交
4525 4526
}

4527
void
4528
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4529 4530 4531
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4532
	nn->reclaim_str_hashtbl_size--;
4533 4534
}

4535
void
4536
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4537 4538 4539 4540 4541
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4542 4543
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4544
			                struct nfs4_client_reclaim, cr_strhash);
4545
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4546 4547
		}
	}
4548
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4549 4550 4551 4552
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4553
struct nfs4_client_reclaim *
4554
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4555 4556 4557 4558
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4559
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4560

4561
	strhashval = clientstr_hashval(recdir);
4562
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4563
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4564 4565 4566 4567 4568 4569 4570 4571 4572
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4573
__be32
4574
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4575
{
4576 4577 4578
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4579
	clp = find_confirmed_client(clid, sessions, nn);
4580 4581 4582 4583
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4584 4585
}

B
Bryan Schumaker 已提交
4586 4587
#ifdef CONFIG_NFSD_FAULT_INJECTION

4588 4589 4590 4591 4592 4593
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
	expire_client(clp);
	return 1;
}

4594 4595 4596
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
4597
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4598 4599 4600 4601 4602 4603 4604 4605
	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];
4606
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4607 4608 4609
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

B
Bryan Schumaker 已提交
4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635
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);
}

4636 4637 4638 4639 4640 4641 4642
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;
}

4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662
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);
}

4663 4664 4665 4666 4667 4668 4669
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;
}

4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715
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;
}

4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727
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;
}

4728
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4729 4730
{
	struct nfs4_client *clp, *next;
4731
	u64 count = 0;
4732
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4733

4734 4735 4736
	if (!nfsd_netns_ready(nn))
		return 0;

4737
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4738 4739
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4740 4741 4742
			break;
	}

4743 4744 4745
	return count;
}

4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760
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
4761 4762
#endif /* CONFIG_NFSD_FAULT_INJECTION */

4763
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4764

4765
void
4766
nfs4_state_init(void)
L
Linus Torvalds 已提交
4767
{
4768
	int i;
L
Linus Torvalds 已提交
4769 4770 4771 4772 4773

	for (i = 0; i < FILE_HASH_SIZE; i++) {
		INIT_LIST_HEAD(&file_hashtbl[i]);
	}
	INIT_LIST_HEAD(&del_recall_lru);
4774 4775
}

4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796
/*
 * 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);
}

4797
static int nfs4_state_create_net(struct net *net)
4798 4799 4800 4801 4802 4803 4804
{
	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)
4805
		goto err;
4806 4807 4808 4809
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4810 4811 4812 4813
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4814 4815 4816 4817
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4818 4819 4820 4821
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4822

4823
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4824
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4825
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4826
	}
4827 4828
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4829 4830
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4831 4832
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4833
	nn->conf_name_tree = RB_ROOT;
4834
	nn->unconf_name_tree = RB_ROOT;
4835
	INIT_LIST_HEAD(&nn->client_lru);
4836
	INIT_LIST_HEAD(&nn->close_lru);
4837
	spin_lock_init(&nn->client_lock);
4838

4839
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4840
	get_net(net);
4841

4842
	return 0;
4843

4844 4845
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4846 4847
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4848 4849
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4850 4851
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4852 4853
err:
	return -ENOMEM;
4854 4855 4856
}

static void
4857
nfs4_state_destroy_net(struct net *net)
4858 4859 4860 4861
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4862
	struct rb_node *node, *tmp;
4863 4864 4865 4866 4867 4868 4869

	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);
		}
	}
4870 4871 4872 4873 4874 4875 4876 4877 4878 4879

	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);
	}

4880
	kfree(nn->sessionid_hashtbl);
4881
	kfree(nn->lockowner_ino_hashtbl);
4882
	kfree(nn->ownerstr_hashtbl);
4883
	kfree(nn->unconf_id_hashtbl);
4884
	kfree(nn->conf_id_hashtbl);
4885
	put_net(net);
4886 4887
}

4888
int
4889
nfs4_state_start_net(struct net *net)
4890
{
4891
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4892 4893
	int ret;

4894
	ret = nfs4_state_create_net(net);
4895 4896
	if (ret)
		return ret;
4897
	nfsd4_client_tracking_init(net);
4898
	nn->boot_time = get_seconds();
4899
	locks_start_grace(net, &nn->nfsd4_manager);
4900
	nn->grace_ended = false;
4901
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4902 4903
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4904 4905 4906 4907 4908 4909 4910 4911 4912 4913
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

4914
	ret = set_callback_cred();
4915 4916
	if (ret)
		return -ENOMEM;
4917
	laundry_wq = create_singlethread_workqueue("nfsd4");
4918 4919 4920 4921
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4922 4923 4924
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4925

4926
	set_max_delegations();
4927

4928
	return 0;
4929

4930 4931
out_free_laundry:
	destroy_workqueue(laundry_wq);
4932
out_recovery:
4933
	return ret;
L
Linus Torvalds 已提交
4934 4935
}

4936
/* should be called with the state lock held */
4937
void
4938
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4939 4940 4941
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4942
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4943

4944 4945
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4946

L
Linus Torvalds 已提交
4947 4948 4949 4950
	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);
4951 4952
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4953 4954 4955 4956 4957 4958 4959 4960
		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);
	}

4961
	nfsd4_client_tracking_exit(net);
4962
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
4963 4964 4965 4966 4967
}

void
nfs4_state_shutdown(void)
{
4968
	destroy_workqueue(laundry_wq);
4969
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
4970
}
4971 4972 4973 4974

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4975 4976
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
4977 4978 4979 4980 4981
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4982 4983 4984 4985 4986 4987 4988 4989 4990 4991
	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);
4992 4993
}

4994 4995 4996
/*
 * functions to set current state id
 */
4997 4998 4999 5000 5001 5002
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5003 5004 5005 5006 5007 5008
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023
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
 */
5024

5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036
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);
}

5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048
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);
}

5049 5050 5051 5052 5053 5054 5055
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5056
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5057
{
5058
	get_stateid(cstate, &locku->lu_stateid);
5059
}
5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071

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);
}