nfs4state.c 132.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 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
241 242 243 244 245 246 247 248
	stid = kmem_cache_alloc(slab, GFP_KERNEL);
	if (!stid)
		return NULL;

	if (!idr_pre_get(stateids, GFP_KERNEL))
		goto out_free;
	if (idr_get_new_above(stateids, stid, min_stateid, &new_id))
		goto out_free;
249
	stid->sc_client = cl;
J
J. Bruce Fields 已提交
250 251 252
	stid->sc_type = 0;
	stid->sc_stateid.si_opaque.so_id = new_id;
	stid->sc_stateid.si_opaque.so_clid = cl->cl_clientid;
253
	/* Will be incremented before return to client: */
J
J. Bruce Fields 已提交
254
	stid->sc_stateid.si_generation = 0;
255 256

	/*
J
J. Bruce Fields 已提交
257 258 259 260 261 262 263
	 * 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):
264
	 */
J
J. Bruce Fields 已提交
265 266 267 268 269 270 271 272

	min_stateid = new_id+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return stid;
out_free:
	kfree(stid);
	return NULL;
273 274
}

275 276 277 278 279
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
280
static struct nfs4_delegation *
281
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
282 283 284 285 286
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
287 288 289 290 291 292 293
	/*
	 * 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;
294 295
	if (fp->fi_had_conflict)
		return NULL;
296
	if (num_delegations > max_delegations)
297
		return NULL;
298
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
299
	if (dp == NULL)
L
Linus Torvalds 已提交
300
		return dp;
J
J. Bruce Fields 已提交
301
	dp->dl_stid.sc_type = NFS4_DELEG_STID;
302 303
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
304 305
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
306 307
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
308
	num_delegations++;
309 310
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
311
	INIT_LIST_HEAD(&dp->dl_recall_lru);
312
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
313 314
	dp->dl_file = fp;
	dp->dl_type = type;
315
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
316 317
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
318
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
319 320 321
	return dp;
}

F
Fengguang Wu 已提交
322
static void free_stid(struct nfs4_stid *s, struct kmem_cache *slab)
J
J. Bruce Fields 已提交
323 324 325 326 327 328 329
{
	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
330 331 332 333 334
void
nfs4_put_delegation(struct nfs4_delegation *dp)
{
	if (atomic_dec_and_test(&dp->dl_count)) {
		dprintk("NFSD: freeing dp %p\n",dp);
335
		put_nfs4_file(dp->dl_file);
J
J. Bruce Fields 已提交
336
		free_stid(&dp->dl_stid, deleg_slab);
337
		num_delegations--;
L
Linus Torvalds 已提交
338 339 340
	}
}

341
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
342
{
343 344 345
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
346
		fput(fp->fi_deleg_file);
347 348
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
349 350
}

J
J. Bruce Fields 已提交
351 352
static void unhash_stid(struct nfs4_stid *s)
{
J
J. Bruce Fields 已提交
353
	s->sc_type = 0;
J
J. Bruce Fields 已提交
354 355
}

L
Linus Torvalds 已提交
356 357 358 359
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
360
	unhash_stid(&dp->dl_stid);
361
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
362
	spin_lock(&recall_lock);
363
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
364 365
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
366
	nfs4_put_deleg_lease(dp->dl_file);
L
Linus Torvalds 已提交
367 368 369 370 371 372 373
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

374 375 376 377 378 379 380 381 382 383
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;
}

384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401
/*
 * 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.
 */
402 403
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
404
	int i;
405
	unsigned int access = 0;
406 407 408

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
409
			access |= i;
410
	}
411
	return access;
412 413
}

414
static bool
415
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
416 417
	unsigned int access, deny;

418 419
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
420
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
421 422
		return false;
	return true;
423 424
}

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445
/* 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);
}

446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
/* 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);
465 466 467 468
}

static int nfs4_access_to_omode(u32 access)
{
469
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
470 471 472 473 474 475 476
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
477 478
	WARN_ON_ONCE(1);
	return O_RDONLY;
479 480
}

481 482 483 484 485 486 487 488 489 490 491 492 493 494
/* 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);
	}
}

495
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
496 497 498 499 500
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

501
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
502
{
503
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
504
	put_nfs4_file(stp->st_file);
505 506 507
	stp->st_file = NULL;
}

508
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
509
{
J
J. Bruce Fields 已提交
510
	free_stid(&stp->st_stid, stateid_slab);
511 512
}

513
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
514 515 516 517
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
518
	unhash_stid(&stp->st_stid);
519 520
	file = find_any_file(stp->st_file);
	if (file)
521
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
522
	close_generic_stateid(stp);
523 524 525
	free_generic_stateid(stp);
}

526
static void unhash_lockowner(struct nfs4_lockowner *lo)
527
{
528
	struct nfs4_ol_stateid *stp;
529

530 531
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
532
	list_del(&lo->lo_owner_ino_hash);
533 534
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
535
				struct nfs4_ol_stateid, st_perstateowner);
536 537 538 539
		release_lock_stateid(stp);
	}
}

540
static void release_lockowner(struct nfs4_lockowner *lo)
541
{
542 543
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
544 545 546
}

static void
547
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
548
{
549
	struct nfs4_lockowner *lo;
550 551

	while (!list_empty(&open_stp->st_lockowners)) {
552 553 554
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
555 556 557
	}
}

558
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
559 560 561
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
562 563 564 565 566 567
	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
568
	unhash_stid(&stp->st_stid);
569 570 571
	free_generic_stateid(stp);
}

572
static void unhash_openowner(struct nfs4_openowner *oo)
573
{
574
	struct nfs4_ol_stateid *stp;
575

576 577 578 579
	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,
580
				struct nfs4_ol_stateid, st_perstateowner);
581
		release_open_stateid(stp);
582 583 584
	}
}

585 586 587 588 589
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
590
		unhash_stid(&s->st_stid);
591 592 593 594 595
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

596
static void release_openowner(struct nfs4_openowner *oo)
597
{
598 599
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
600
	release_last_closed_stateid(oo);
601
	nfs4_free_openowner(oo);
602 603
}

M
Marc Eshel 已提交
604 605 606 607 608 609 610 611
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

612
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
613 614 615 616 617 618
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]);
}
619 620 621 622 623 624 625
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

M
Marc Eshel 已提交
626

A
Andy Adamson 已提交
627 628 629 630 631 632 633 634 635 636 637 638 639
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;
}

/*
640 641 642 643 644 645 646 647 648 649 650 651 652
 * 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)

653 654 655 656 657 658 659 660 661
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]);
}

662
/*
663 664 665 666 667 668 669 670
 * 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;
}

671
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
672
{
673 674
	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
675

676 677
	return size;
}
A
Andy Adamson 已提交
678

679 680
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
681
 * re-negotiate active sessions and reduce their slot usage to make
682
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
683
 */
684
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
685
{
686
	int avail;
A
Andy Adamson 已提交
687

688
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
689

690
	spin_lock(&nfsd_drc_lock);
691 692
	avail = min((unsigned long)NFSD_MAX_MEM_PER_SESSION,
		    nfsd_drc_max_mem - nfsd_drc_mem_used);
693 694 695
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
696

697 698
	return num;
}
A
Andy Adamson 已提交
699

700 701
static void nfsd4_put_drc_mem(int slotsize, int num)
{
702
	spin_lock(&nfsd_drc_lock);
703
	nfsd_drc_mem_used -= slotsize * num;
704
	spin_unlock(&nfsd_drc_lock);
705
}
A
Andy Adamson 已提交
706

707
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
708 709 710
{
	struct nfsd4_session *new;
	int mem, i;
711

712 713 714
	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
715

716 717 718
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
719
	/* allocate each struct nfsd4_slot and data cache in one piece */
720 721 722 723
	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])
724 725
			goto out_free;
	}
726 727 728 729 730 731
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
732 733
}

734 735 736 737
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
738
{
739
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
740

741
	new->maxreqs = numslots;
742 743
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
744 745 746 747
	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
748

749 750 751 752 753
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
754

755 756 757 758
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
759

760 761 762 763 764 765
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
	spin_unlock(&clp->cl_lock);
766
	nfsd4_probe_callback(clp);
767
}
A
Andy Adamson 已提交
768

769
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
770 771
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
772

773 774
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
775
		return NULL;
776 777
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
778
	conn->cn_flags = flags;
779 780 781
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
782

783 784 785 786
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
787 788
}

789
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
790
{
791
	struct nfs4_client *clp = ses->se_client;
792

793
	spin_lock(&clp->cl_lock);
794
	__nfsd4_hash_conn(conn, ses);
795
	spin_unlock(&clp->cl_lock);
796 797
}

798
static int nfsd4_register_conn(struct nfsd4_conn *conn)
799
{
800
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
801
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
802 803
}

804
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
805
{
806
	int ret;
A
Andy Adamson 已提交
807

808
	nfsd4_hash_conn(conn, ses);
809 810 811 812
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
813
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
814 815 816
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
817
}
A
Andy Adamson 已提交
818

819
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
820 821 822
{
	u32 dir = NFS4_CDFC4_FORE;

823
	if (cses->flags & SESSION4_BACK_CHAN)
824
		dir |= NFS4_CDFC4_BACK;
825
	return alloc_conn(rqstp, dir);
826 827 828
}

/* must be called under client_lock */
829
static void nfsd4_del_conns(struct nfsd4_session *s)
830
{
831 832
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
833

834 835 836 837 838
	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);
839

840 841
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
842

843 844 845
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
846
}
A
Andy Adamson 已提交
847

848 849 850 851 852 853 854
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);
}

855
static void free_session(struct kref *kref)
856 857
{
	struct nfsd4_session *ses;
858
	struct nfsd_net *nn;
859 860

	ses = container_of(kref, struct nfsd4_session, se_ref);
861 862 863
	nn = net_generic(ses->se_client->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
864
	nfsd4_del_conns(ses);
865
	__free_session(ses);
866 867
}

868 869
void nfsd4_put_session(struct nfsd4_session *ses)
{
870 871 872
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
873
	nfsd4_put_session_locked(ses);
874
	spin_unlock(&nn->client_lock);
875 876
}

877 878
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan,
					   struct nfsd_net *nn)
879 880 881 882 883 884 885 886 887 888 889 890
{
	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);
891 892
	if (numslots < 1)
		return NULL;
893

894
	new = __alloc_session(slotsize, numslots);
895
	if (!new) {
896
		nfsd4_put_drc_mem(slotsize, numslots);
897
		return NULL;
898
	}
899
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize, nn);
900 901
	return new;
}
902

903
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
904 905
{
	int idx;
906
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
907

A
Andy Adamson 已提交
908 909 910
	new->se_client = clp;
	gen_sessionid(new);

911 912
	INIT_LIST_HEAD(&new->se_conns);

913
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
914
	new->se_flags = cses->flags;
915
	new->se_cb_prog = cses->callback_prog;
916
	new->se_cb_sec = cses->cb_sec;
A
Andy Adamson 已提交
917
	kref_init(&new->se_ref);
918
	idx = hash_sessionid(&new->se_sessionid);
919
	spin_lock(&nn->client_lock);
920
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
921
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
922
	list_add(&new->se_perclnt, &clp->cl_sessions);
923
	spin_unlock(&clp->cl_lock);
924
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
925

926
	if (cses->flags & SESSION4_BACK_CHAN) {
927
		struct sockaddr *sa = svc_addr(rqstp);
928 929 930 931 932 933 934
		/*
		 * 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:
		 */
935 936 937
		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
938 939
}

940
/* caller must hold client_lock */
M
Marc Eshel 已提交
941
static struct nfsd4_session *
942
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
943 944 945
{
	struct nfsd4_session *elem;
	int idx;
946
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
947 948 949 950

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
951
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
952 953 954 955 956 957 958 959 960 961
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

962
/* caller must hold client_lock */
A
Andy Adamson 已提交
963
static void
M
Marc Eshel 已提交
964
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
965 966
{
	list_del(&ses->se_hash);
967
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
968
	list_del(&ses->se_perclnt);
969
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
970 971
}

972
/* must be called under the client_lock */
L
Linus Torvalds 已提交
973
static inline void
974
renew_client_locked(struct nfs4_client *clp)
L
Linus Torvalds 已提交
975
{
976 977
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

B
Benny Halevy 已提交
978
	if (is_client_expired(clp)) {
979 980
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
B
Benny Halevy 已提交
981 982 983 984 985 986
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

L
Linus Torvalds 已提交
987 988 989
	dprintk("renewing client (clientid %08x/%08x)\n", 
			clp->cl_clientid.cl_boot, 
			clp->cl_clientid.cl_id);
990
	list_move_tail(&clp->cl_lru, &nn->client_lru);
L
Linus Torvalds 已提交
991 992 993
	clp->cl_time = get_seconds();
}

994 995 996
static inline void
renew_client(struct nfs4_client *clp)
{
997 998 999
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
1000
	renew_client_locked(clp);
1001
	spin_unlock(&nn->client_lock);
1002 1003
}

L
Linus Torvalds 已提交
1004 1005
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1006
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1007
{
1008
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1009
		return 0;
A
Andy Adamson 已提交
1010
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1011
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1012 1013 1014 1015 1016 1017 1018 1019
	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.
 */
1020
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1021 1022 1023
{
	struct nfs4_client *clp;

1024 1025 1026
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1027
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1028 1029 1030
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1031
	}
1032
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1033 1034 1035 1036 1037 1038
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1039
	struct nfsd_net __maybe_unused *nn = net_generic(clp->net, nfsd_net_id);
1040 1041

	lockdep_assert_held(&nn->client_lock);
1042 1043 1044 1045 1046
	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);
1047
		nfsd4_put_session_locked(ses);
1048
	}
1049
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1050
	kfree(clp->cl_name.data);
M
majianpeng 已提交
1051
	idr_destroy(&clp->cl_stateids);
L
Linus Torvalds 已提交
1052 1053 1054
	kfree(clp);
}

1055 1056 1057 1058
void
release_session_client(struct nfsd4_session *session)
{
	struct nfs4_client *clp = session->se_client;
1059
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
1060

1061
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
1062 1063 1064 1065 1066 1067
		return;
	if (is_client_expired(clp)) {
		free_client(clp);
		session->se_client = NULL;
	} else
		renew_client_locked(clp);
1068
	spin_unlock(&nn->client_lock);
1069 1070
}

B
Benny Halevy 已提交
1071 1072 1073 1074
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1075 1076
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1077
	mark_client_expired(clp);
B
Benny Halevy 已提交
1078
	list_del(&clp->cl_lru);
1079
	spin_lock(&clp->cl_lock);
1080 1081
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1082
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1083 1084
}

L
Linus Torvalds 已提交
1085
static void
1086
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1087
{
1088
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1089 1090
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1091
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1092 1093 1094

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1095 1096 1097
	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
1098 1099 1100 1101 1102 1103 1104
		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);
	}
1105
	while (!list_empty(&clp->cl_openowners)) {
1106 1107
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1108
	}
1109
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1110 1111
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1112
	list_del(&clp->cl_idhash);
1113
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1114
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1115
	else
1116
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1117
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1118
	unhash_client_locked(clp);
1119 1120
	if (atomic_read(&clp->cl_refcount) == 0)
		free_client(clp);
1121
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1122 1123
}

1124 1125 1126 1127 1128 1129
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1130 1131 1132 1133
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
1134 1135
}

1136 1137
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1138 1139 1140 1141
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1142
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1143
{
1144 1145 1146 1147 1148 1149 1150
	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;
1151
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1152 1153 1154 1155
	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);
1156
	return 0;
L
Linus Torvalds 已提交
1157 1158
}

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169
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);
}

1170
static int same_name(const char *n1, const char *n2)
1171
{
N
NeilBrown 已提交
1172
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1173 1174 1175
}

static int
1176 1177 1178
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1179 1180 1181
}

static int
1182 1183 1184
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
1185 1186
}

1187 1188 1189 1190 1191 1192 1193
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++)
1194
		if (!gid_eq(GROUP_AT(g1, i), GROUP_AT(g2, i)))
1195 1196 1197 1198
			return false;
	return true;
}

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214
/*
 * 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);
}


1215
static bool
1216 1217
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1218
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1219 1220
		|| (!uid_eq(cr1->cr_uid, cr2->cr_uid))
		|| (!gid_eq(cr1->cr_gid, cr2->cr_gid))
1221 1222 1223 1224 1225 1226
		|| !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;
1227
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1228 1229
}

1230
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1231 1232 1233
{
	static u32 current_clientid = 1;

1234
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1235 1236 1237
	clp->cl_clientid.cl_id = current_clientid++; 
}

1238 1239
static void gen_confirm(struct nfs4_client *clp)
{
1240
	__be32 verf[2];
1241
	static u32 i;
L
Linus Torvalds 已提交
1242

1243 1244 1245
	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
1246 1247
}

1248
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1249
{
J
J. Bruce Fields 已提交
1250 1251 1252 1253 1254 1255
	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
1256 1257
}

1258
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1259 1260
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1261

1262
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1263 1264
	if (!s)
		return NULL;
1265
	if (typemask & s->sc_type)
1266 1267 1268 1269
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1270
static struct nfs4_client *create_client(struct xdr_netobj name,
1271 1272 1273 1274
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1275
	int ret;
1276
	struct net *net = SVC_NET(rqstp);
1277
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1278 1279 1280 1281 1282

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

1283
	INIT_LIST_HEAD(&clp->cl_sessions);
1284 1285
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1286
		spin_lock(&nn->client_lock);
1287
		free_client(clp);
1288
		spin_unlock(&nn->client_lock);
1289
		return NULL;
1290
	}
1291
	idr_init(&clp->cl_stateids);
1292
	atomic_set(&clp->cl_refcount, 0);
1293
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1294 1295 1296 1297
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1298
	INIT_LIST_HEAD(&clp->cl_callbacks);
1299
	spin_lock_init(&clp->cl_lock);
1300
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1301
	clp->cl_time = get_seconds();
1302 1303 1304 1305 1306
	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);
1307
	clp->cl_cb_session = NULL;
1308
	clp->net = net;
1309 1310 1311
	return clp;
}

1312
static void
1313 1314 1315 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
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
1354 1355
{
	unsigned int idhashval;
1356
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1357

1358
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1359
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1360
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1361
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1362
	renew_client(clp);
L
Linus Torvalds 已提交
1363 1364
}

1365
static void
L
Linus Torvalds 已提交
1366 1367 1368
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1369
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1370 1371

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1372
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1373
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1374
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1375
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1376 1377 1378 1379
	renew_client(clp);
}

static struct nfs4_client *
1380
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1381 1382 1383 1384
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1385
	list_for_each_entry(clp, &nn->conf_id_hashtbl[idhashval], cl_idhash) {
1386
		if (same_clid(&clp->cl_clientid, clid)) {
1387 1388
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1389
			renew_client(clp);
L
Linus Torvalds 已提交
1390
			return clp;
1391
		}
L
Linus Torvalds 已提交
1392 1393 1394 1395 1396
	}
	return NULL;
}

static struct nfs4_client *
1397
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1398 1399 1400 1401
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1402
	list_for_each_entry(clp, &nn->unconf_id_hashtbl[idhashval], cl_idhash) {
1403 1404 1405
		if (same_clid(&clp->cl_clientid, clid)) {
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
L
Linus Torvalds 已提交
1406
			return clp;
1407
		}
L
Linus Torvalds 已提交
1408 1409 1410 1411
	}
	return NULL;
}

1412
static bool clp_used_exchangeid(struct nfs4_client *clp)
1413
{
1414
	return clp->cl_exchange_flags != 0;
1415
} 
1416

1417
static struct nfs4_client *
1418
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1419
{
1420
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1421 1422 1423
}

static struct nfs4_client *
1424
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1425
{
1426
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1427 1428
}

1429
static void
1430
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1431
{
1432
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1433 1434
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
	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
1445 1446
		goto out_err;

1447
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1448
					    se->se_callback_addr_len,
1449 1450
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1451

1452
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1453
		goto out_err;
1454

1455 1456
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1457

1458 1459
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1460
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1461 1462
	return;
out_err:
1463 1464
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1465
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1466 1467 1468 1469 1470 1471
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1472
/*
1473
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1474 1475 1476 1477
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1478 1479
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1480

1481
	dprintk("--> %s slot %p\n", __func__, slot);
1482

1483 1484
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1485

1486
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1487
	if (nfsd4_not_cached(resp)) {
1488
		slot->sl_datalen = 0;
1489
		return;
1490
	}
1491 1492 1493 1494 1495 1496 1497
	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;
1498 1499 1500
}

/*
1501 1502 1503 1504
 * 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.
1505 1506
 *
 */
1507 1508 1509
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1510
{
1511 1512
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1513

1514 1515 1516
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1517

1518
	/* Return nfserr_retry_uncached_rep in next operation. */
1519
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1520 1521 1522
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1523
	}
1524
	return op->status;
1525 1526 1527
}

/*
1528 1529
 * The sequence operation is not cached because we can use the slot and
 * session values.
1530 1531
 */
__be32
1532 1533
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1534
{
1535
	struct nfsd4_slot *slot = resp->cstate.slot;
1536 1537
	__be32 status;

1538
	dprintk("--> %s slot %p\n", __func__, slot);
1539

1540 1541 1542 1543
	/* 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;
1544

1545 1546
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1547

1548 1549 1550
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1551 1552 1553 1554

	return status;
}

A
Andy Adamson 已提交
1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
/*
 * 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;
}

1571 1572 1573 1574 1575 1576 1577 1578
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.
	 */
1579 1580 1581
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1582 1583
}

A
Andy Adamson 已提交
1584 1585 1586 1587 1588
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1589
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1590
	__be32 status;
1591
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1592
	nfs4_verifier		verf = exid->verifier;
1593
	struct sockaddr		*sa = svc_addr(rqstp);
1594
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1595
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1596

1597
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1598
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1599
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1600
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1601
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1602

1603
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1604 1605 1606 1607 1608 1609
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1610 1611
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1612 1613 1614 1615 1616
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1617
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1618
	nfs4_lock_state();
1619
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1620
	if (conf) {
1621 1622 1623
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1624 1625
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1626
				status = nfserr_inval;
1627 1628
				goto out;
			}
1629
			if (!creds_match) { /* case 9 */
1630
				status = nfserr_perm;
1631 1632 1633
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1634 1635 1636
				status = nfserr_not_same;
				goto out;
			}
1637 1638 1639 1640
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1641
		}
1642
		if (!creds_match) { /* case 3 */
1643 1644
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1645 1646 1647 1648 1649
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1650
		if (verfs_match) { /* case 2 */
1651
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1652 1653 1654 1655 1656
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1657 1658
	}

1659
	if (update) { /* case 7 */
1660 1661
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1662 1663
	}

1664
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1665
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1666 1667
		expire_client(unconf);

1668
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1669
out_new:
J
Jeff Layton 已提交
1670
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1671
	if (new == NULL) {
1672
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1673 1674
		goto out;
	}
1675
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1676

1677
	gen_clid(new, nn);
1678
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1679 1680 1681 1682
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1683
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1684 1685 1686
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1687
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1688 1689 1690 1691 1692
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1693 1694
}

J
J. Bruce Fields 已提交
1695
static __be32
1696
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1697
{
1698 1699
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1700 1701

	/* The slot is in use, and no response has been sent. */
1702 1703
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1704 1705 1706 1707
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1708
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1709
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1710
		return nfs_ok;
1711
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1712 1713 1714 1715
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1716 1717 1718 1719 1720 1721 1722
/*
 * 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
1723
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
{
	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;
}

1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
#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
1754
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1755 1756 1757 1758 1759
{
	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
1760 1761 1762 1763 1764
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1765
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1766
	struct nfs4_client *conf, *unconf;
1767
	struct nfsd4_session *new;
1768
	struct nfsd4_conn *conn;
1769
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1770
	__be32 status = 0;
1771
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1772

1773 1774
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1775 1776
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1777
	new = alloc_session(&cr_ses->fore_channel, nn);
1778 1779 1780 1781 1782 1783
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1784

A
Andy Adamson 已提交
1785
	nfs4_lock_state();
1786
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1787
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
A
Andy Adamson 已提交
1788 1789

	if (conf) {
1790 1791
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1792
		if (status == nfserr_replay_cache) {
1793
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1794
			goto out_free_conn;
1795
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1796
			status = nfserr_seq_misordered;
1797
			goto out_free_conn;
A
Andy Adamson 已提交
1798 1799
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1800
		struct nfs4_client *old;
A
Andy Adamson 已提交
1801
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1802
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1803
			status = nfserr_clid_inuse;
1804
			goto out_free_conn;
A
Andy Adamson 已提交
1805
		}
1806 1807
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1808 1809
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1810
			status = nfserr_seq_misordered;
1811
			goto out_free_conn;
A
Andy Adamson 已提交
1812
		}
1813
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
J
J. Bruce Fields 已提交
1814 1815 1816
		if (old)
			expire_client(old);
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1817 1818 1819
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1820
		goto out_free_conn;
A
Andy Adamson 已提交
1821
	}
1822
	status = nfs_ok;
1823 1824 1825 1826 1827 1828
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1829 1830 1831
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1832
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1833
	       NFS4_MAX_SESSIONID_LEN);
1834 1835
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1836
	cs_slot->sl_seqid++;
1837
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1838

1839 1840
	/* 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
1841
	nfs4_unlock_state();
1842
out:
A
Andy Adamson 已提交
1843 1844
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
1845
out_free_conn:
1846
	nfs4_unlock_state();
1847 1848 1849 1850
	free_conn(conn);
out_free_session:
	__free_session(new);
	goto out;
A
Andy Adamson 已提交
1851 1852
}

1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866
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;
}

1867 1868 1869
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1870
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1871

1872
	spin_lock(&nn->client_lock);
1873 1874
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1875
	spin_unlock(&nn->client_lock);
1876 1877 1878 1879 1880 1881

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1882 1883 1884 1885 1886
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1887
	struct nfsd4_conn *conn;
1888
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1889 1890 1891

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1892
	spin_lock(&nn->client_lock);
1893
	cstate->session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1894 1895 1896 1897 1898 1899
	/* 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);
	}
1900
	spin_unlock(&nn->client_lock);
1901 1902 1903 1904
	if (!cstate->session)
		return nfserr_badsession;

	status = nfsd4_map_bcts_dir(&bcts->dir);
1905 1906 1907 1908 1909
	if (status)
		return status;
	conn = alloc_conn(rqstp, bcts->dir);
	if (!conn)
		return nfserr_jukebox;
1910
	nfsd4_init_conn(rqstp, conn, cstate->session);
1911
	return nfs_ok;
1912 1913
}

1914 1915 1916 1917 1918 1919 1920
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
1921 1922 1923 1924 1925
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1926
	struct nfsd4_session *ses;
J
J. Bruce Fields 已提交
1927
	__be32 status = nfserr_badsession;
1928
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1929 1930 1931 1932 1933 1934 1935 1936

	/* 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?
	 */

1937
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1938 1939 1940
		if (!nfsd4_last_compound_op(r))
			return nfserr_not_only_op;
	}
B
Benny Halevy 已提交
1941
	dump_sessionid(__func__, &sessionid->sessionid);
1942
	spin_lock(&nn->client_lock);
1943
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
B
Benny Halevy 已提交
1944
	if (!ses) {
1945
		spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1946 1947 1948 1949
		goto out;
	}

	unhash_session(ses);
1950
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1951

1952
	nfs4_lock_state();
1953
	nfsd4_probe_callback_sync(ses->se_client);
1954
	nfs4_unlock_state();
1955

1956
	spin_lock(&nn->client_lock);
1957
	nfsd4_del_conns(ses);
1958
	nfsd4_put_session_locked(ses);
1959
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1960 1961 1962 1963
	status = nfs_ok;
out:
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
1964 1965
}

1966
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
1967 1968 1969 1970
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
1971
		if (c->cn_xprt == xpt) {
1972 1973 1974 1975 1976 1977
			return c;
		}
	}
	return NULL;
}

1978
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
1979 1980
{
	struct nfs4_client *clp = ses->se_client;
1981
	struct nfsd4_conn *c;
1982
	int ret;
1983 1984

	spin_lock(&clp->cl_lock);
1985
	c = __nfsd4_find_conn(new->cn_xprt, ses);
1986 1987 1988 1989 1990 1991 1992
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
1993 1994 1995 1996
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
1997 1998 1999
	return;
}

2000 2001 2002 2003 2004 2005 2006
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
2007 2008 2009 2010 2011 2012 2013 2014
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
2015
__be32
B
Benny Halevy 已提交
2016
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2017 2018 2019
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2020
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2021 2022
	struct nfsd4_session *session;
	struct nfsd4_slot *slot;
2023
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2024
	__be32 status;
2025
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2026

2027 2028 2029
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2030 2031 2032 2033 2034 2035 2036 2037
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2038
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2039
	status = nfserr_badsession;
2040
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2041 2042 2043
	if (!session)
		goto out;

2044 2045 2046 2047
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
		goto out;

M
Mi Jinlong 已提交
2048 2049 2050 2051
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
		goto out;

B
Benny Halevy 已提交
2052
	status = nfserr_badslot;
2053
	if (seq->slotid >= session->se_fchannel.maxreqs)
B
Benny Halevy 已提交
2054 2055
		goto out;

2056
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2057 2058
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2059 2060 2061 2062 2063
	/* 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;

2064 2065
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2066
	if (status == nfserr_replay_cache) {
2067 2068 2069
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
			goto out;
B
Benny Halevy 已提交
2070 2071
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2072
		/* Return the cached reply status and set cstate->status
2073
		 * for nfsd4_proc_compound processing */
2074
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2075
		cstate->status = nfserr_replay_cache;
2076
		goto out;
B
Benny Halevy 已提交
2077 2078 2079 2080
	}
	if (status)
		goto out;

2081 2082
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2083

B
Benny Halevy 已提交
2084 2085
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2086
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2087 2088
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2089 2090
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2091 2092 2093 2094 2095

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

out:
2096
	/* Hold a session reference until done processing the compound. */
2097
	if (cstate->session) {
2098 2099
		struct nfs4_client *clp = session->se_client;

2100
		nfsd4_get_session(cstate->session);
2101
		atomic_inc(&clp->cl_refcount);
2102 2103
		switch (clp->cl_cb_state) {
		case NFSD4_CB_DOWN:
2104
			seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
2105 2106
			break;
		case NFSD4_CB_FAULT:
2107
			seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
2108
			break;
2109 2110
		default:
			seq->status_flags = 0;
2111
		}
2112
	}
2113
	kfree(conn);
2114
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2115 2116
	dprintk("%s: return %d\n", __func__, ntohl(status));
	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 2130 2131 2132

	if (conf) {
		clp = conf;

2133
		if (!is_client_expired(conf) && client_has_state(conf)) {
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
			status = nfserr_clientid_busy;
			goto out;
		}

		/* rfc5661 18.50.3 */
		if (cstate->session && conf == cstate->session->se_client) {
			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();
	dprintk("%s return %d\n", __func__, ntohl(status));
	return status;
}

2157 2158 2159
__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
2160
	__be32 status = 0;
2161

2162 2163 2164 2165 2166 2167 2168 2169 2170
	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;
	}
2171

2172
	nfs4_lock_state();
2173
	status = nfserr_complete_already;
2174 2175
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2176 2177 2178 2179
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2180 2181 2182 2183 2184 2185 2186
		/*
		 * 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.
		 */
2187 2188 2189
		goto out;

	status = nfs_ok;
2190
	nfsd4_client_record_create(cstate->session->se_client);
2191
out:
2192
	nfs4_unlock_state();
2193
	return status;
2194 2195
}

2196
__be32
2197 2198
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2199
{
2200
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2201
	nfs4_verifier		clverifier = setclid->se_verf;
2202
	struct nfs4_client	*conf, *unconf, *new;
2203
	__be32 			status;
2204 2205
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2206
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2207
	nfs4_lock_state();
2208
	conf = find_confirmed_client_by_name(&clname, nn);
2209
	if (conf) {
2210
		/* case 0: */
L
Linus Torvalds 已提交
2211
		status = nfserr_clid_inuse;
2212 2213
		if (clp_used_exchangeid(conf))
			goto out;
2214
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2215 2216 2217 2218 2219
			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
2220 2221 2222
			goto out;
		}
	}
2223
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2224 2225
	if (unconf)
		expire_client(unconf);
2226
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2227
	new = create_client(clname, rqstp, &clverifier);
2228 2229
	if (new == NULL)
		goto out;
2230
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2231
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2232
		copy_clid(new, conf);
2233
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2234
		gen_clid(new, nn);
2235
	new->cl_minorversion = 0;
2236
	gen_callback(new, setclid, rqstp);
2237
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2238 2239 2240 2241 2242 2243 2244 2245 2246 2247
	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;
}


2248
__be32
2249 2250 2251
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2252
{
2253
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2254 2255
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2256
	__be32 status;
2257
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2258

2259
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2260 2261
		return nfserr_stale_clientid;
	nfs4_lock_state();
2262

2263
	conf = find_confirmed_client(clid, false, nn);
2264
	unconf = find_unconfirmed_client(clid, false, nn);
2265
	/*
2266 2267 2268 2269
	 * 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.
2270
	 */
2271 2272 2273 2274 2275
	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;
2276
	/* cases below refer to rfc 3530 section 14.2.34: */
2277 2278
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2279
			status = nfs_ok;
2280 2281 2282 2283 2284 2285
		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 */
2286 2287 2288
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2289
	} else { /* case 3: normal case; new or rebooted client */
2290
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2291
		if (conf)
2292 2293
			expire_client(conf);
		move_to_confirmed(unconf);
2294
		nfsd4_probe_callback(unconf);
2295
	}
L
Linus Torvalds 已提交
2296 2297 2298 2299 2300
out:
	nfs4_unlock_state();
	return status;
}

2301 2302 2303 2304 2305
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2306
/* OPEN Share state helper functions */
2307
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2308 2309 2310
{
	unsigned int hashval = file_hashval(ino);

2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
	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
2323 2324
}

N
NeilBrown 已提交
2325
static void
2326
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2327
{
N
NeilBrown 已提交
2328 2329
	if (*slab == NULL)
		return;
2330
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2331
	*slab = NULL;
L
Linus Torvalds 已提交
2332 2333
}

2334
void
L
Linus Torvalds 已提交
2335 2336
nfsd4_free_slabs(void)
{
2337 2338
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2339
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2340
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2341
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2342
}
L
Linus Torvalds 已提交
2343

2344
int
N
NeilBrown 已提交
2345 2346
nfsd4_init_slabs(void)
{
2347 2348 2349 2350 2351
	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",
2352
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2353
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2354 2355
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2356
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2357 2358
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2359
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2360
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2361 2362
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2363
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2364
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2365 2366
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2367 2368 2369 2370 2371
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2372 2373
}

2374 2375 2376 2377 2378 2379 2380
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
2381
{
2382 2383
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2384 2385
}

2386
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2387
{
2388 2389 2390
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2391 2392
}

2393
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2394
{
L
Linus Torvalds 已提交
2395 2396
	struct nfs4_stateowner *sop;

2397
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2398 2399 2400 2401 2402
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2403
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2404
		return NULL;
2405 2406 2407
	}
	sop->so_owner.len = owner->len;

2408
	INIT_LIST_HEAD(&sop->so_stateids);
2409 2410 2411 2412 2413
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2414
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2415
{
2416 2417 2418
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2419
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2420 2421
}

2422
static struct nfs4_openowner *
2423
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2424
	struct nfs4_openowner *oo;
2425

2426 2427
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2428
		return NULL;
2429 2430
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2431
	oo->oo_flags = NFS4_OO_NEW;
2432
	oo->oo_time = 0;
2433
	oo->oo_last_closed_stid = NULL;
2434 2435 2436
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2437 2438
}

2439
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2440
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2441

J
J. Bruce Fields 已提交
2442
	stp->st_stid.sc_type = NFS4_OPEN_STID;
2443
	INIT_LIST_HEAD(&stp->st_lockowners);
2444
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2445
	list_add(&stp->st_perfile, &fp->fi_stateids);
2446
	stp->st_stateowner = &oo->oo_owner;
2447
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2448 2449 2450
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2451
	set_access(open->op_share_access, stp);
2452
	set_deny(open->op_share_deny, stp);
2453
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2454 2455
}

2456
static void
2457
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2458
{
2459 2460
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2461
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2462

2463
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2464
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2465 2466 2467
}

static int
2468 2469 2470 2471 2472 2473
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
2474 2475
}

2476
static struct nfs4_openowner *
2477 2478
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2479
{
2480 2481
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2482
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2483

2484
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2485 2486
		if (!so->so_is_open_owner)
			continue;
2487 2488
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2489 2490 2491
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2492 2493 2494
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505
	}
	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;

2506
	spin_lock(&recall_lock);
L
Linus Torvalds 已提交
2507
	list_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2508 2509
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2510
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2511
			return fp;
2512
		}
L
Linus Torvalds 已提交
2513
	}
2514
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2515 2516 2517 2518 2519 2520 2521
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2522
static __be32
L
Linus Torvalds 已提交
2523 2524 2525 2526
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;
2527
	struct nfs4_ol_stateid *stp;
2528
	__be32 ret;
L
Linus Torvalds 已提交
2529 2530 2531 2532

	dprintk("NFSD: nfs4_share_conflict\n");

	fp = find_file(ino);
2533 2534
	if (!fp)
		return nfs_ok;
2535
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2536
	/* Search for conflicting share reservations */
2537
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2538 2539
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2540
			goto out;
L
Linus Torvalds 已提交
2541
	}
2542 2543 2544 2545
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2546 2547
}

2548
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558
{
	/* 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);

2559
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2560 2561
	dp->dl_time = get_seconds();

2562 2563 2564
	nfsd4_cb_recall(dp);
}

2565
/* Called from break_lease() with lock_flocks() held. */
2566 2567
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2568 2569
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2570

2571 2572 2573 2574 2575 2576 2577 2578
	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;
	}
2579 2580
	/*
	 * We don't want the locks code to timeout the lease for us;
2581
	 * we'll remove it ourself if a delegation isn't returned
2582
	 * in time:
2583 2584
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2585

2586
	spin_lock(&recall_lock);
2587 2588 2589
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2590
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
}

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

2602
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2603 2604
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2605 2606
};

2607 2608 2609 2610 2611 2612 2613 2614 2615 2616
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
2617

2618
__be32
A
Andy Adamson 已提交
2619
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2620
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2621 2622 2623 2624
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2625
	struct nfs4_openowner *oo = NULL;
2626
	__be32 status;
L
Linus Torvalds 已提交
2627

2628
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2629
		return nfserr_stale_clientid;
2630 2631 2632 2633 2634 2635 2636
	/*
	 * 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
2637

2638
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2639
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2640 2641
	open->op_openowner = oo;
	if (!oo) {
2642 2643
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2644
		if (clp == NULL)
2645
			return nfserr_expired;
2646
		goto new_owner;
L
Linus Torvalds 已提交
2647
	}
2648
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2649
		/* Replace unconfirmed owners without checking for replay. */
2650 2651 2652
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2653
		goto new_owner;
2654
	}
2655 2656 2657 2658 2659
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2660 2661 2662 2663 2664
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2665 2666 2667 2668
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2669
	return nfs_ok;
L
Linus Torvalds 已提交
2670 2671
}

2672
static inline __be32
N
NeilBrown 已提交
2673 2674 2675 2676 2677 2678 2679 2680
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;
}

2681
static int share_access_to_flags(u32 share_access)
2682
{
2683
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2684 2685
}

2686
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2687
{
2688
	struct nfs4_stid *ret;
2689

2690
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2691 2692 2693
	if (!ret)
		return NULL;
	return delegstateid(ret);
2694 2695
}

2696 2697 2698 2699 2700 2701
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;
}

2702
static __be32
2703
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2704 2705 2706
		struct nfs4_delegation **dp)
{
	int flags;
2707
	__be32 status = nfserr_bad_stateid;
2708

2709
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2710
	if (*dp == NULL)
2711
		goto out;
2712
	flags = share_access_to_flags(open->op_share_access);
2713 2714 2715
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2716
out:
2717
	if (!nfsd4_is_deleg_cur(open))
2718 2719 2720
		return nfs_ok;
	if (status)
		return status;
2721
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2722
	return nfs_ok;
2723 2724
}

2725
static __be32
2726
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2727
{
2728
	struct nfs4_ol_stateid *local;
2729
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2730

2731
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2732 2733 2734 2735
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2736
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2737 2738 2739
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2740
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2741
	}
2742
	return nfs_ok;
L
Linus Torvalds 已提交
2743 2744
}

2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755
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;
}

2756 2757
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2758 2759
{
	__be32 status;
2760 2761 2762
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773
	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;
}

2774
static inline __be32
L
Linus Torvalds 已提交
2775 2776 2777 2778 2779 2780 2781 2782 2783 2784
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))
2785
		return nfserr_inval;
L
Linus Torvalds 已提交
2786 2787 2788
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2789
static __be32
2790
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
2791
{
2792
	u32 op_share_access = open->op_share_access;
2793
	bool new_access;
2794
	__be32 status;
L
Linus Torvalds 已提交
2795

2796
	new_access = !test_access(op_share_access, stp);
2797
	if (new_access) {
2798
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2799 2800
		if (status)
			return status;
2801
	}
L
Linus Torvalds 已提交
2802 2803
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2804
		if (new_access) {
2805
			int oflag = nfs4_access_to_omode(op_share_access);
2806 2807
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2808 2809 2810
		return status;
	}
	/* remember the open */
2811
	set_access(op_share_access, stp);
2812
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2813 2814 2815 2816 2817 2818

	return nfs_ok;
}


static void
2819
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2820
{
2821
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2822 2823
}

2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836
/* 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;
}

2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848
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;
2849
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2850 2851 2852 2853
	fl->fl_pid = current->tgid;
	return fl;
}

2854 2855
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2856
	struct nfs4_file *fp = dp->dl_file;
2857 2858 2859 2860 2861 2862
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2863
	fl->fl_file = find_readable_file(fp);
2864
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2865
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2866
	if (status) {
2867
		list_del_init(&dp->dl_perclnt);
2868 2869 2870
		locks_free_lock(fl);
		return -ENOMEM;
	}
2871
	fp->fi_lease = fl;
A
Al Viro 已提交
2872
	fp->fi_deleg_file = get_file(fl->fl_file);
2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891
	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);
2892
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2893 2894 2895
	return 0;
}

2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911
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:
2912
			WARN_ON_ONCE(1);
2913 2914 2915 2916
		}
	}
}

L
Linus Torvalds 已提交
2917 2918 2919 2920
/*
 * Attempt to hand out a delegation.
 */
static void
2921 2922
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
2923 2924
{
	struct nfs4_delegation *dp;
2925
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2926
	int cb_up;
2927
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2928

2929
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2930
	flag = NFS4_OPEN_DELEGATE_NONE;
2931 2932 2933
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2934
			if (!cb_up)
2935 2936 2937 2938 2939 2940 2941 2942
				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.... */
2943
			if (locks_in_grace(net))
2944
				goto out;
2945
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2946 2947 2948 2949 2950 2951 2952 2953 2954
				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
2955

2956
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2957 2958
	if (dp == NULL)
		goto out_no_deleg;
2959
	status = nfs4_set_delegation(dp, flag);
2960
	if (status)
2961
		goto out_free;
L
Linus Torvalds 已提交
2962

2963
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2964

2965
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
2966
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2967 2968
out:
	open->op_delegate_type = flag;
2969 2970 2971 2972 2973
	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");

2974 2975 2976
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
2977
	}
2978 2979
	return;
out_free:
2980
	unhash_stid(&dp->dl_stid);
2981
	nfs4_put_delegation(dp);
2982 2983 2984
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
2985 2986
}

2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004
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
3005 3006 3007
/*
 * called with nfs4_lock_state() held.
 */
3008
__be32
L
Linus Torvalds 已提交
3009 3010
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3011
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3012
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3013 3014
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3015
	struct nfs4_ol_stateid *stp = NULL;
3016
	struct nfs4_delegation *dp = NULL;
3017
	__be32 status;
L
Linus Torvalds 已提交
3018 3019 3020 3021 3022 3023 3024 3025 3026 3027

	/*
	 * 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;
3028
		status = nfs4_check_deleg(cl, fp, open, &dp);
3029 3030
		if (status)
			goto out;
L
Linus Torvalds 已提交
3031
	} else {
3032
		status = nfserr_bad_stateid;
3033
		if (nfsd4_is_deleg_cur(open))
3034
			goto out;
3035
		status = nfserr_jukebox;
3036 3037 3038
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3039 3040 3041 3042 3043 3044 3045 3046
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3047
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3048 3049 3050
		if (status)
			goto out;
	} else {
3051
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3052 3053 3054
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3055
		if (status)
L
Linus Torvalds 已提交
3056
			goto out;
3057 3058
		stp = open->op_stp;
		open->op_stp = NULL;
3059
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3060
	}
3061 3062
	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
3063

3064
	if (nfsd4_has_session(&resp->cstate)) {
3065
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3066

3067 3068 3069 3070 3071 3072 3073
		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
3074 3075 3076 3077
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3078
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3079
nodeleg:
L
Linus Torvalds 已提交
3080 3081
	status = nfs_ok;

3082
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3083
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3084
out:
3085 3086
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3087 3088
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3089

3090 3091
	if (fp)
		put_nfs4_file(fp);
3092
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3093
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3094 3095 3096 3097
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3098
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3099
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3100 3101 3102 3103 3104
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119
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;
		}
	}
3120 3121
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3122
	if (open->op_stp)
3123
		free_generic_stateid(open->op_stp);
3124 3125
}

3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137
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;
}

3138
__be32
3139 3140
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3141 3142
{
	struct nfs4_client *clp;
3143
	__be32 status;
3144
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3145 3146 3147 3148

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3149 3150
	status = lookup_clientid(clid, cstate->minorversion, nn, &clp);
	if (status)
L
Linus Torvalds 已提交
3151 3152
		goto out;
	status = nfserr_cb_path_down;
3153
	if (!list_empty(&clp->cl_delegations)
3154
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3155 3156 3157 3158 3159 3160 3161
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3162
static void
3163
nfsd4_end_grace(struct nfsd_net *nn)
3164
{
3165
	/* do nothing if grace period already ended */
3166
	if (nn->grace_ended)
3167 3168
		return;

3169
	dprintk("NFSD: end of grace period\n");
3170
	nn->grace_ended = true;
3171
	nfsd4_record_grace_done(nn, nn->boot_time);
3172
	locks_end_grace(&nn->nfsd4_manager);
3173 3174 3175 3176 3177
	/*
	 * 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:
	 */
3178
	nn->nfsd4_grace = nn->nfsd4_lease;
3179 3180
}

3181
static time_t
3182
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3183 3184
{
	struct nfs4_client *clp;
3185
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3186 3187
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3188 3189 3190
	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
3191 3192 3193 3194

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3195
	nfsd4_end_grace(nn);
3196
	INIT_LIST_HEAD(&reaplist);
3197
	spin_lock(&nn->client_lock);
3198
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3199 3200 3201 3202 3203 3204 3205
		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;
		}
3206 3207 3208 3209 3210 3211 3212
		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);
3213
	}
3214
	spin_unlock(&nn->client_lock);
3215 3216
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3217 3218 3219 3220 3221 3222 3223
		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);
3224 3225
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238
		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);
	}
3239
	test_val = nn->nfsd4_lease;
3240
	list_for_each_safe(pos, next, &nn->close_lru) {
3241 3242 3243
		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
3244 3245 3246 3247
			if (test_val > u)
				test_val = u;
			break;
		}
3248
		release_openowner(oo);
L
Linus Torvalds 已提交
3249 3250 3251 3252 3253 3254 3255
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3256 3257 3258 3259
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3260
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3261 3262
{
	time_t t;
3263 3264 3265 3266
	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
3267

3268
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3269
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3270
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3271 3272
}

3273
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3274
{
3275 3276 3277
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3278 3279 3280
}

static int
3281
STALE_STATEID(stateid_t *stateid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3282
{
3283
	if (stateid->si_opaque.so_clid.cl_boot == nn->boot_time)
3284 3285
		return 0;
	dprintk("NFSD: stale stateid " STATEID_FMT "!\n",
3286
		STATEID_VAL(stateid));
3287
	return 1;
L
Linus Torvalds 已提交
3288 3289 3290
}

static inline int
3291
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3292
{
3293 3294 3295
	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
3296 3297 3298
}

static inline int
3299
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3300
{
3301 3302
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3303 3304 3305
}

static
3306
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3307
{
3308
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3309

3310 3311 3312
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3313
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3314
                goto out;
3315
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3316 3317 3318 3319 3320 3321
                goto out;
	status = nfs_ok;
out:
	return status;
}

3322
static inline __be32
3323
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3324
{
3325
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3326
		return nfs_ok;
3327
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3328
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343
		 * 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
3344
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3345
{
3346
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3347 3348
}

3349 3350 3351 3352 3353 3354
/* 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
3355
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3356
{
A
Andy Adamson 已提交
3357 3358 3359 3360
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3361
	if (has_session && in->si_generation == 0)
3362 3363 3364 3365
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3366

3367
	/* If the client sends us a stateid from the future, it's buggy: */
3368
	if (stateid_generation_after(in, ref))
3369 3370
		return nfserr_bad_stateid;
	/*
3371 3372 3373 3374 3375 3376 3377 3378
	 * 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:
3379
	 */
3380
	return nfserr_old_stateid;
3381 3382
}

3383
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3384
{
3385 3386 3387
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3388

3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399
	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;
	}
3400
	s = find_stateid(cl, stateid);
3401
	if (!s)
3402
		return nfserr_bad_stateid;
3403
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3404
	if (status)
3405 3406 3407 3408 3409
		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
3410
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3411 3412
		return nfserr_bad_stateid;
	return nfs_ok;
3413 3414
}

3415 3416 3417
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3418 3419 3420 3421 3422
{
	struct nfs4_client *cl;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
3423
	if (STALE_STATEID(stateid, nn))
3424
		return nfserr_stale_stateid;
3425
	cl = find_confirmed_client(&stateid->si_opaque.so_clid, sessions, nn);
3426 3427 3428 3429 3430 3431 3432 3433 3434
	if (!cl)
		return nfserr_expired;
	*s = find_stateid_by_type(cl, stateid, typemask);
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;

}

L
Linus Torvalds 已提交
3435 3436 3437
/*
* Checks for stateid operations
*/
3438
__be32
3439
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3440
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3441
{
3442
	struct nfs4_stid *s;
3443
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3444
	struct nfs4_delegation *dp = NULL;
3445
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3446
	struct inode *ino = current_fh->fh_dentry->d_inode;
3447
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3448
	__be32 status;
L
Linus Torvalds 已提交
3449 3450 3451 3452

	if (filpp)
		*filpp = NULL;

3453
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3454 3455 3456
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3457
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3458

3459 3460
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3461 3462
	if (status)
		return status;
3463 3464 3465
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3466 3467
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3468
		dp = delegstateid(s);
3469 3470 3471
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3472
		if (filpp) {
3473
			*filpp = dp->dl_file->fi_deleg_file;
3474 3475 3476 3477 3478
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3479
		}
3480 3481 3482
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3483
		stp = openlockstateid(s);
3484 3485
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3486
			goto out;
3487
		if (stp->st_stateowner->so_is_open_owner
3488
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3489
			goto out;
3490 3491
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3492
			goto out;
3493 3494 3495 3496 3497 3498
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3499 3500 3501
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3502 3503 3504 3505 3506 3507
	}
	status = nfs_ok;
out:
	return status;
}

3508
static __be32
3509
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3510
{
3511
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3512 3513 3514 3515 3516
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3517 3518 3519 3520 3521 3522 3523
/*
 * 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)
{
3524 3525 3526 3527 3528
	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)
3529 3530
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3531 3532
	nfs4_unlock_state();

3533 3534 3535
	return nfs_ok;
}

3536 3537 3538 3539 3540
__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
3541
	struct nfs4_stid *s;
3542
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3543
	__be32 ret = nfserr_bad_stateid;
3544 3545

	nfs4_lock_state();
3546
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3547
	if (!s)
3548
		goto out;
J
J. Bruce Fields 已提交
3549 3550
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3551 3552
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3553 3554 3555 3556 3557 3558 3559 3560 3561
	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;
3562 3563 3564
		break;
	default:
		ret = nfserr_bad_stateid;
3565 3566 3567 3568 3569 3570
	}
out:
	nfs4_unlock_state();
	return ret;
}

3571 3572 3573 3574 3575 3576
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
3577

3578
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3579 3580 3581 3582 3583 3584 3585 3586
{
	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;
3587 3588 3589 3590 3591 3592 3593 3594 3595 3596
	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);
3597 3598
}

L
Linus Torvalds 已提交
3599 3600 3601
/* 
 * Checks for sequence id mutating operations. 
 */
3602
static __be32
3603
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3604
			 stateid_t *stateid, char typemask,
3605 3606
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3607
{
3608
	__be32 status;
3609
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3610

3611 3612
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3613

L
Linus Torvalds 已提交
3614
	*stpp = NULL;
3615 3616
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3617 3618
	if (status)
		return status;
3619
	*stpp = openlockstateid(s);
3620
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3621

3622 3623
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3624

3625 3626
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)
3627 3628 3629
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3630

3631
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3632
						NFS4_OPEN_STID, stpp, nn);
3633 3634
	if (status)
		return status;
3635
	oo = openowner((*stpp)->st_stateowner);
3636
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3637 3638
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3639 3640
}

3641
__be32
3642
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3643
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3644
{
3645
	__be32 status;
3646
	struct nfs4_openowner *oo;
3647
	struct nfs4_ol_stateid *stp;
3648
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3649 3650

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3651 3652
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3653

3654
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3655 3656
	if (status)
		return status;
L
Linus Torvalds 已提交
3657 3658 3659

	nfs4_lock_state();

3660
	status = nfs4_preprocess_seqid_op(cstate,
3661
					oc->oc_seqid, &oc->oc_req_stateid,
3662
					NFS4_OPEN_STID, &stp, nn);
3663
	if (status)
3664
		goto out;
3665
	oo = openowner(stp->st_stateowner);
3666
	status = nfserr_bad_stateid;
3667
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3668
		goto out;
3669
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3670 3671
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3672
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3673
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3674

3675
	nfsd4_client_record_create(oo->oo_owner.so_client);
3676
	status = nfs_ok;
L
Linus Torvalds 已提交
3677
out:
3678 3679
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3680 3681 3682
	return status;
}

J
J. Bruce Fields 已提交
3683
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3684
{
3685
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3686 3687
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3688
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3689
}
3690

J
J. Bruce Fields 已提交
3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704
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:
3705
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3706 3707 3708 3709
	}
}

static void
3710
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3711 3712 3713 3714
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3715
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3716 3717 3718
	}
}

3719
__be32
3720 3721
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3722
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3723
{
3724
	__be32 status;
3725
	struct nfs4_ol_stateid *stp;
3726
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3727 3728

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3729 3730
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3731

3732
	/* We don't yet support WANT bits: */
3733 3734 3735
	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
3736 3737

	nfs4_lock_state();
3738
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3739
					&od->od_stateid, &stp, nn);
3740
	if (status)
L
Linus Torvalds 已提交
3741 3742
		goto out; 
	status = nfserr_inval;
3743 3744
	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
3745 3746 3747
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3748
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3749 3750 3751 3752
		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
3753
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3754

3755
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3756

3757 3758
	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
3759 3760
	status = nfs_ok;
out:
3761 3762
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3763 3764 3765
	return status;
}

3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782
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;
3783 3784 3785 3786 3787 3788 3789
	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;
3790 3791
}

L
Linus Torvalds 已提交
3792 3793 3794
/*
 * nfs4_unlock_state() called after encode
 */
3795
__be32
3796
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3797
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3798
{
3799
	__be32 status;
3800
	struct nfs4_openowner *oo;
3801
	struct nfs4_ol_stateid *stp;
3802 3803
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3804 3805

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3806 3807
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3808 3809

	nfs4_lock_state();
3810 3811 3812
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3813
					&stp, nn);
3814
	if (status)
L
Linus Torvalds 已提交
3815
		goto out; 
3816
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3817
	status = nfs_ok;
3818 3819
	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
3820

3821
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3822
	release_last_closed_stateid(oo);
3823
	oo->oo_last_closed_stid = stp;
3824

3825 3826 3827 3828 3829 3830 3831 3832 3833 3834
	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))
3835
				move_to_close_lru(oo, SVC_NET(rqstp));
3836 3837
		}
	}
L
Linus Torvalds 已提交
3838
out:
3839 3840
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3841 3842 3843
	return status;
}

3844
__be32
3845 3846
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3847
{
3848 3849
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3850
	struct nfs4_stid *s;
3851
	__be32 status;
3852
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3853

3854
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3855
		return status;
L
Linus Torvalds 已提交
3856 3857

	nfs4_lock_state();
3858 3859
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3860
	if (status)
3861
		goto out;
3862
	dp = delegstateid(s);
3863
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3864 3865 3866 3867
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3868
out:
3869 3870
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3871 3872 3873 3874 3875 3876
	return status;
}


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

3877
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3878

B
Benny Halevy 已提交
3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893
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;

3894
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3895 3896 3897 3898
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3899
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3900 3901 3902
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3903
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922
}

/*
 * 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;
}

3923 3924
/* 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. */
3925
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3926
};
L
Linus Torvalds 已提交
3927 3928 3929 3930

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3931
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3932

3933
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3934 3935 3936
		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);
3937 3938 3939
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3940 3941
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3942
	} else {
3943 3944 3945
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3946 3947
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3948 3949
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3950 3951
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3952 3953 3954 3955 3956 3957
		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
3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968
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;
}

3969 3970
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3971
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3972
{
3973
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
3974
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3975

3976
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
3977 3978
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
3979 3980 3981 3982
	}
	return NULL;
}

3983
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
3984
{
3985 3986 3987
	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);
3988
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
3989

3990
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
3991
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
3992
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
3993 3994
}

L
Linus Torvalds 已提交
3995 3996 3997
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
3998
 * occurred. 
L
Linus Torvalds 已提交
3999
 *
4000
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4001 4002
 */

4003
static struct nfs4_lockowner *
4004
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4005
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4006

4007 4008
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4009
		return NULL;
4010 4011
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4012 4013
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4014 4015 4016
	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
4017 4018
}

4019 4020
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
4021
{
4022
	struct nfs4_ol_stateid *stp;
4023
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4024

4025
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4026
	if (stp == NULL)
J
J. Bruce Fields 已提交
4027
		return NULL;
J
J. Bruce Fields 已提交
4028
	stp->st_stid.sc_type = NFS4_LOCK_STID;
4029
	list_add(&stp->st_perfile, &fp->fi_stateids);
4030 4031
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4032
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4033
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4034
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4035
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4036
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4037 4038 4039
	return stp;
}

4040
static int
L
Linus Torvalds 已提交
4041 4042
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4043
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4044 4045 4046
	     LOFF_OVERFLOW(offset, length)));
}

4047
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4048 4049 4050 4051
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4052
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4053 4054
		return;
	nfs4_file_get_access(fp, oflag);
4055
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4056 4057
}

J
J. Bruce Fields 已提交
4058
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)
4059 4060 4061 4062 4063 4064
{
	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;
4065
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4066

4067 4068
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4069 4070 4071 4072 4073 4074 4075 4076
	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;
	}
4077
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090
			&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
4091 4092 4093
/*
 *  LOCK operation 
 */
4094
__be32
4095
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4096
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4097
{
4098 4099
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4100
	struct nfs4_ol_stateid *lock_stp;
4101
	struct file *filp = NULL;
4102 4103
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4104
	__be32 status = 0;
4105
	bool new_state = false;
4106
	int lkflg;
4107
	int err;
4108 4109
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4110 4111 4112 4113 4114 4115 4116 4117

	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;

4118
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4119
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4120 4121 4122 4123
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4124 4125 4126
	nfs4_lock_state();

	if (lock->lk_is_new) {
4127
		struct nfs4_ol_stateid *open_stp = NULL;
4128 4129 4130 4131 4132 4133 4134

		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
4135
		status = nfserr_stale_clientid;
4136
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4137 4138 4139
			goto out;

		/* validate and update open stateid and open seqid */
4140
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4141 4142
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4143
					&open_stp, nn);
4144
		if (status)
L
Linus Torvalds 已提交
4145
			goto out;
4146
		open_sop = openowner(open_stp->st_stateowner);
4147
		status = nfserr_bad_stateid;
4148
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4149 4150
						&lock->v.new.clientid))
			goto out;
4151 4152
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4153
	} else
4154
		status = nfs4_preprocess_seqid_op(cstate,
4155 4156
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4157
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4158 4159
	if (status)
		goto out;
4160
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4161

4162 4163 4164 4165 4166
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4167
	status = nfserr_grace;
4168
	if (locks_in_grace(net) && !lock->lk_reclaim)
4169 4170
		goto out;
	status = nfserr_no_grace;
4171
	if (!locks_in_grace(net) && lock->lk_reclaim)
4172 4173
		goto out;

4174 4175 4176 4177 4178 4179 4180 4181
	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
4182 4183 4184
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4185 4186 4187
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4188
			file_lock->fl_type = F_RDLCK;
4189
			break;
L
Linus Torvalds 已提交
4190 4191
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4192 4193 4194
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4195
			file_lock->fl_type = F_WRLCK;
4196
			break;
L
Linus Torvalds 已提交
4197 4198 4199 4200
		default:
			status = nfserr_inval;
		goto out;
	}
4201 4202 4203 4204
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219
	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
4220

4221
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4222
	switch (-err) {
L
Linus Torvalds 已提交
4223
	case 0: /* success! */
4224 4225
		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
4226
				sizeof(stateid_t));
4227
		status = 0;
4228 4229 4230 4231
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4232
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4233
		break;
L
Linus Torvalds 已提交
4234 4235
	case (EDEADLK):
		status = nfserr_deadlock;
4236
		break;
4237
	default:
4238
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4239
		status = nfserrno(err);
4240
		break;
L
Linus Torvalds 已提交
4241 4242
	}
out:
4243
	if (status && new_state)
4244
		release_lockowner(lock_sop);
4245 4246
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4247 4248 4249 4250
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4251 4252 4253
	return status;
}

4254 4255 4256 4257 4258 4259
/*
 * 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.)
 */
4260
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4261 4262
{
	struct file *file;
4263 4264 4265 4266 4267
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4268 4269 4270
	return err;
}

L
Linus Torvalds 已提交
4271 4272 4273
/*
 * LOCKT operation
 */
4274
__be32
4275 4276
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4277 4278
{
	struct inode *inode;
4279
	struct file_lock *file_lock = NULL;
4280
	struct nfs4_lockowner *lo;
4281
	__be32 status;
4282
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4283

4284
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4285 4286 4287 4288 4289 4290 4291
		return nfserr_grace;

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

	nfs4_lock_state();

4292 4293 4294 4295 4296
	if (!nfsd4_has_session(cstate)) {
		status = lookup_clientid(&lockt->lt_clientid, false, nn, NULL);
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
4297

4298
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4299 4300
		goto out;

4301
	inode = cstate->current_fh.fh_dentry->d_inode;
4302 4303 4304 4305 4306 4307 4308
	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
4309 4310 4311
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4312
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4313 4314 4315
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4316
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4317 4318
		break;
		default:
4319
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4320 4321 4322 4323
			status = nfserr_inval;
		goto out;
	}

4324
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4325
	if (lo)
4326 4327 4328
		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
4329

4330 4331
	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
4332

4333
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4334

4335
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4336
	if (status)
4337
		goto out;
4338

4339
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4340
		status = nfserr_denied;
4341
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4342 4343 4344
	}
out:
	nfs4_unlock_state();
4345 4346
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4347 4348 4349
	return status;
}

4350
__be32
4351
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4352
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4353
{
4354
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4355
	struct file *filp = NULL;
4356
	struct file_lock *file_lock = NULL;
4357
	__be32 status;
4358
	int err;
4359 4360
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4361 4362 4363 4364 4365 4366 4367 4368 4369
	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();
									        
4370
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4371 4372
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4373
	if (status)
L
Linus Torvalds 已提交
4374
		goto out;
4375 4376 4377 4378 4379
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397
	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
4398 4399 4400 4401

	/*
	*  Try to unlock the file in the VFS.
	*/
4402
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4403
	if (err) {
4404
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4405 4406 4407 4408 4409
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4410 4411
	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
4412 4413

out:
4414 4415
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4416 4417
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4418 4419 4420
	return status;

out_nfserr:
4421
	status = nfserrno(err);
L
Linus Torvalds 已提交
4422 4423 4424 4425 4426 4427 4428 4429 4430
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4431
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4432 4433
{
	struct file_lock **flpp;
4434
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4435 4436
	int status = 0;

4437
	lock_flocks();
L
Linus Torvalds 已提交
4438
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4439
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4440 4441
			status = 1;
			goto out;
4442
		}
L
Linus Torvalds 已提交
4443 4444
	}
out:
4445
	unlock_flocks();
L
Linus Torvalds 已提交
4446 4447 4448
	return status;
}

4449
__be32
4450 4451 4452
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4453 4454
{
	clientid_t *clid = &rlockowner->rl_clientid;
4455
	struct nfs4_stateowner *sop;
4456
	struct nfs4_lockowner *lo;
4457
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4458
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4459
	struct list_head matches;
4460
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4461
	__be32 status;
4462
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4463 4464 4465 4466 4467 4468

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

	nfs4_lock_state();

4469 4470 4471 4472
	status = lookup_clientid(clid, cstate->minorversion, nn, NULL);
	if (status)
		goto out;

4473 4474
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4475

4476
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4477 4478
		if (sop->so_is_open_owner)
			continue;
4479 4480 4481 4482 4483 4484 4485 4486
		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
4487
		}
4488 4489 4490 4491 4492
	}
	/* 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
4493
	while (!list_empty(&matches)) {
4494 4495
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4496 4497
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4498 4499
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4500 4501 4502 4503 4504 4505 4506
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4507
alloc_reclaim(void)
L
Linus Torvalds 已提交
4508
{
N
NeilBrown 已提交
4509
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4510 4511
}

4512
bool
4513
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4514
{
4515
	struct nfs4_client_reclaim *crp;
4516

4517
	crp = nfsd4_find_reclaim_client(name, nn);
4518
	return (crp && crp->cr_clp);
4519 4520
}

L
Linus Torvalds 已提交
4521 4522 4523
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4524
struct nfs4_client_reclaim *
4525
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4526 4527
{
	unsigned int strhashval;
4528
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4529

N
NeilBrown 已提交
4530 4531
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4532 4533 4534
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4535
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4536
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4537
		crp->cr_clp = NULL;
4538
		nn->reclaim_str_hashtbl_size++;
4539 4540
	}
	return crp;
L
Linus Torvalds 已提交
4541 4542
}

4543
void
4544
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4545 4546 4547
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4548
	nn->reclaim_str_hashtbl_size--;
4549 4550
}

4551
void
4552
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4553 4554 4555 4556 4557
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4558 4559
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4560
			                struct nfs4_client_reclaim, cr_strhash);
4561
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4562 4563
		}
	}
4564
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4565 4566 4567 4568
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4569
struct nfs4_client_reclaim *
4570
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4571 4572 4573 4574
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4575
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4576

4577
	strhashval = clientstr_hashval(recdir);
4578
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4579
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4580 4581 4582 4583 4584 4585 4586 4587 4588
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4589
__be32
4590
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4591
{
4592 4593 4594
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4595
	clp = find_confirmed_client(clid, sessions, nn);
4596 4597 4598 4599
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4600 4601
}

B
Bryan Schumaker 已提交
4602 4603
#ifdef CONFIG_NFSD_FAULT_INJECTION

4604 4605 4606 4607 4608 4609
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
	expire_client(clp);
	return 1;
}

4610 4611 4612
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
4613
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4614 4615 4616 4617 4618 4619 4620 4621
	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];
4622
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4623 4624 4625
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

B
Bryan Schumaker 已提交
4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651
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);
}

4652 4653 4654 4655 4656 4657 4658
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;
}

4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678
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);
}

4679 4680 4681 4682 4683 4684 4685
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;
}

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 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731
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;
}

4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743
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;
}

4744
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4745 4746
{
	struct nfs4_client *clp, *next;
4747
	u64 count = 0;
4748
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4749

4750 4751 4752
	if (!nfsd_netns_ready(nn))
		return 0;

4753
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4754 4755
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4756 4757 4758
			break;
	}

4759 4760 4761
	return count;
}

4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776
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
4777 4778
#endif /* CONFIG_NFSD_FAULT_INJECTION */

4779
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4780

4781
void
4782
nfs4_state_init(void)
L
Linus Torvalds 已提交
4783
{
4784
	int i;
L
Linus Torvalds 已提交
4785 4786 4787 4788 4789

	for (i = 0; i < FILE_HASH_SIZE; i++) {
		INIT_LIST_HEAD(&file_hashtbl[i]);
	}
	INIT_LIST_HEAD(&del_recall_lru);
4790 4791
}

4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812
/*
 * 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);
}

4813
static int nfs4_state_create_net(struct net *net)
4814 4815 4816 4817 4818 4819 4820
{
	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)
4821
		goto err;
4822 4823 4824 4825
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4826 4827 4828 4829
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4830 4831 4832 4833
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4834 4835 4836 4837
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4838

4839
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4840
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4841
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4842
	}
4843 4844
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4845 4846
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4847 4848
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4849
	nn->conf_name_tree = RB_ROOT;
4850
	nn->unconf_name_tree = RB_ROOT;
4851
	INIT_LIST_HEAD(&nn->client_lru);
4852
	INIT_LIST_HEAD(&nn->close_lru);
4853
	spin_lock_init(&nn->client_lock);
4854

4855
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4856
	get_net(net);
4857

4858
	return 0;
4859

4860 4861
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4862 4863
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4864 4865
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4866 4867
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4868 4869
err:
	return -ENOMEM;
4870 4871 4872
}

static void
4873
nfs4_state_destroy_net(struct net *net)
4874 4875 4876 4877
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4878
	struct rb_node *node, *tmp;
4879 4880 4881 4882 4883 4884 4885

	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);
		}
	}
4886 4887 4888 4889 4890 4891 4892 4893 4894 4895

	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);
	}

4896
	kfree(nn->sessionid_hashtbl);
4897
	kfree(nn->lockowner_ino_hashtbl);
4898
	kfree(nn->ownerstr_hashtbl);
4899
	kfree(nn->unconf_id_hashtbl);
4900
	kfree(nn->conf_id_hashtbl);
4901
	put_net(net);
4902 4903
}

4904
int
4905
nfs4_state_start_net(struct net *net)
4906
{
4907
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4908 4909
	int ret;

4910
	ret = nfs4_state_create_net(net);
4911 4912
	if (ret)
		return ret;
4913
	nfsd4_client_tracking_init(net);
4914
	nn->boot_time = get_seconds();
4915
	locks_start_grace(net, &nn->nfsd4_manager);
4916
	nn->grace_ended = false;
4917
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4918 4919
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4920 4921 4922 4923 4924 4925 4926 4927 4928 4929
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

4930
	ret = set_callback_cred();
4931 4932
	if (ret)
		return -ENOMEM;
4933
	laundry_wq = create_singlethread_workqueue("nfsd4");
4934 4935 4936 4937
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4938 4939 4940
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4941

4942
	set_max_delegations();
4943

4944
	return 0;
4945

4946 4947
out_free_laundry:
	destroy_workqueue(laundry_wq);
4948
out_recovery:
4949
	return ret;
L
Linus Torvalds 已提交
4950 4951
}

4952
/* should be called with the state lock held */
4953
void
4954
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4955 4956 4957
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4958
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4959

4960 4961
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4962

L
Linus Torvalds 已提交
4963 4964 4965 4966
	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);
4967 4968
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4969 4970 4971 4972 4973 4974 4975 4976
		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);
	}

4977
	nfsd4_client_tracking_exit(net);
4978
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
4979 4980 4981 4982 4983
}

void
nfs4_state_shutdown(void)
{
4984
	destroy_workqueue(laundry_wq);
4985
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
4986
}
4987 4988 4989 4990

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4991 4992
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
4993 4994 4995 4996 4997
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4998 4999 5000 5001 5002 5003 5004 5005 5006 5007
	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);
5008 5009
}

5010 5011 5012
/*
 * functions to set current state id
 */
5013 5014 5015 5016 5017 5018
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5019 5020 5021 5022 5023 5024
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039
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
 */
5040

5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052
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);
}

5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064
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);
}

5065 5066 5067 5068 5069 5070 5071
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5072
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5073
{
5074
	get_stateid(cstate, &locku->lu_stateid);
5075
}
5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087

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);
}