nfs4state.c 132.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
/*
*  Copyright (c) 2001 The Regents of the University of Michigan.
*  All rights reserved.
*
*  Kendrick Smith <kmsmith@umich.edu>
*  Andy Adamson <kandros@umich.edu>
*
*  Redistribution and use in source and binary forms, with or without
*  modification, are permitted provided that the following conditions
*  are met:
*
*  1. Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*  2. Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in the
*     documentation and/or other materials provided with the distribution.
*  3. Neither the name of the University nor the names of its
*     contributors may be used to endorse or promote products derived
*     from this software without specific prior written permission.
*
*  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
*  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
*  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
*  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
*  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
*  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
*  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
*  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
*  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
*  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
*  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/

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

48 49
#include "netns.h"

L
Linus Torvalds 已提交
50 51
#define NFSDDBG_FACILITY                NFSDDBG_PROC

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

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

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

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

73 74 75
/* Locking: */

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

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

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

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

97
static void free_session(struct nfsd4_session *);
98

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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


L
Linus Torvalds 已提交
215 216 217 218 219 220 221 222 223 224 225 226 227
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;
}

228 229 230 231 232
static void nfsd4_free_file(struct nfs4_file *f)
{
	kmem_cache_free(file_slab, f);
}

233 234 235
static inline void
put_nfs4_file(struct nfs4_file *fi)
{
236
	if (atomic_dec_and_lock(&fi->fi_ref, &recall_lock)) {
237
		hlist_del(&fi->fi_hash);
238 239
		spin_unlock(&recall_lock);
		iput(fi->fi_inode);
240
		nfsd4_free_file(fi);
241
	}
242 243 244 245 246
}

static inline void
get_nfs4_file(struct nfs4_file *fi)
{
247
	atomic_inc(&fi->fi_ref);
248 249
}

250
static int num_delegations;
251
unsigned long max_delegations;
252 253 254 255 256

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

257 258 259 260
/* 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)
261

262
static unsigned int ownerstr_hashval(u32 clientid, struct xdr_netobj *ownername)
263 264 265 266 267
{
	unsigned int ret;

	ret = opaque_hashval(ownername->data, ownername->len);
	ret += clientid;
268
	return ret & OWNER_HASH_MASK;
269
}
270 271 272 273

/* hash table for nfs4_file */
#define FILE_HASH_BITS                   8
#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
S
Shan Wei 已提交
274

275 276 277 278 279 280
static unsigned int file_hashval(struct inode *ino)
{
	/* XXX: why are we hashing on inode pointer, anyway? */
	return hash_ptr(ino, FILE_HASH_BITS);
}

281
static struct hlist_head file_hashtbl[FILE_HASH_SIZE];
282

283
static void __nfs4_file_get_access(struct nfs4_file *fp, int oflag)
284
{
285
	WARN_ON_ONCE(!(fp->fi_fds[oflag] || fp->fi_fds[O_RDWR]));
286 287 288
	atomic_inc(&fp->fi_access[oflag]);
}

289 290 291 292 293 294 295 296 297 298
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)
299 300 301 302 303 304 305
{
	if (fp->fi_fds[oflag]) {
		fput(fp->fi_fds[oflag]);
		fp->fi_fds[oflag] = NULL;
	}
}

306
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
307 308 309
{
	if (atomic_dec_and_test(&fp->fi_access[oflag])) {
		nfs4_file_put_fd(fp, oflag);
310 311 312 313 314 315 316 317
		/*
		 * 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);
318 319 320
	}
}

321 322 323 324 325 326 327 328 329
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
330 331
static struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl, struct
kmem_cache *slab)
332
{
J
J. Bruce Fields 已提交
333 334 335
	struct idr *stateids = &cl->cl_stateids;
	static int min_stateid = 0;
	struct nfs4_stid *stid;
J
J. Bruce Fields 已提交
336
	int new_id;
337

J
J. Bruce Fields 已提交
338 339 340 341
	stid = kmem_cache_alloc(slab, GFP_KERNEL);
	if (!stid)
		return NULL;

T
Tejun Heo 已提交
342 343
	new_id = idr_alloc(stateids, stid, min_stateid, 0, GFP_KERNEL);
	if (new_id < 0)
J
J. Bruce Fields 已提交
344
		goto out_free;
345
	stid->sc_client = cl;
J
J. Bruce Fields 已提交
346 347 348
	stid->sc_type = 0;
	stid->sc_stateid.si_opaque.so_id = new_id;
	stid->sc_stateid.si_opaque.so_clid = cl->cl_clientid;
349
	/* Will be incremented before return to client: */
J
J. Bruce Fields 已提交
350
	stid->sc_stateid.si_generation = 0;
351 352

	/*
J
J. Bruce Fields 已提交
353 354 355 356 357 358 359
	 * 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):
360
	 */
J
J. Bruce Fields 已提交
361 362 363 364 365 366

	min_stateid = new_id+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return stid;
out_free:
367
	kmem_cache_free(slab, stid);
J
J. Bruce Fields 已提交
368
	return NULL;
369 370
}

371 372 373 374 375
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
376
static struct nfs4_delegation *
377
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
378 379 380 381 382
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
383 384 385 386 387 388 389
	/*
	 * 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;
390 391
	if (fp->fi_had_conflict)
		return NULL;
392
	if (num_delegations > max_delegations)
393
		return NULL;
394
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
395
	if (dp == NULL)
L
Linus Torvalds 已提交
396
		return dp;
J
J. Bruce Fields 已提交
397
	dp->dl_stid.sc_type = NFS4_DELEG_STID;
398 399
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
400 401
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
402 403
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
404
	num_delegations++;
405 406
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
407
	INIT_LIST_HEAD(&dp->dl_recall_lru);
408
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
409 410
	dp->dl_file = fp;
	dp->dl_type = type;
411
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
412 413
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
414
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
415 416 417
	return dp;
}

418
static void remove_stid(struct nfs4_stid *s)
J
J. Bruce Fields 已提交
419 420 421 422 423 424
{
	struct idr *stateids = &s->sc_client->cl_stateids;

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

L
Linus Torvalds 已提交
425 426 427 428
void
nfs4_put_delegation(struct nfs4_delegation *dp)
{
	if (atomic_dec_and_test(&dp->dl_count)) {
429
		kmem_cache_free(deleg_slab, dp);
430
		num_delegations--;
L
Linus Torvalds 已提交
431 432 433
	}
}

434
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
435
{
436 437 438
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
439
		fput(fp->fi_deleg_file);
440 441
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
442 443
}

J
J. Bruce Fields 已提交
444 445
static void unhash_stid(struct nfs4_stid *s)
{
J
J. Bruce Fields 已提交
446
	s->sc_type = 0;
J
J. Bruce Fields 已提交
447 448
}

L
Linus Torvalds 已提交
449 450 451 452
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
453
	unhash_stid(&dp->dl_stid);
454
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
455
	spin_lock(&recall_lock);
456
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
457 458
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
459
	nfs4_put_deleg_lease(dp->dl_file);
460 461 462
	put_nfs4_file(dp->dl_file);
	dp->dl_file = NULL;
	remove_stid(&dp->dl_stid);
L
Linus Torvalds 已提交
463 464 465 466 467 468 469
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

470 471 472 473 474 475 476 477 478 479
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;
}

480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
/*
 * 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.
 */
498 499
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
500
	int i;
501
	unsigned int access = 0;
502 503 504

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
505
			access |= i;
506
	}
507
	return access;
508 509
}

510
static bool
511
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
512 513
	unsigned int access, deny;

514 515
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
516
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
517 518
		return false;
	return true;
519 520
}

521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
/* 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);
}

542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
/* 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);
561 562 563 564
}

static int nfs4_access_to_omode(u32 access)
{
565
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
566 567 568 569 570 571 572
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
573 574
	WARN_ON_ONCE(1);
	return O_RDONLY;
575 576
}

577 578 579 580 581 582 583 584 585 586 587 588 589 590
/* 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);
	}
}

591
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
592 593 594 595 596
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

597
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
598
{
599
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
600
	put_nfs4_file(stp->st_file);
601 602 603
	stp->st_file = NULL;
}

604
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
605
{
606 607
	remove_stid(&stp->st_stid);
	kmem_cache_free(stateid_slab, stp);
608 609
}

610
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
611 612 613 614
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
615
	unhash_stid(&stp->st_stid);
616 617
	file = find_any_file(stp->st_file);
	if (file)
618
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
619
	close_generic_stateid(stp);
620 621 622
	free_generic_stateid(stp);
}

623
static void unhash_lockowner(struct nfs4_lockowner *lo)
624
{
625
	struct nfs4_ol_stateid *stp;
626

627 628
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
629
	list_del(&lo->lo_owner_ino_hash);
630 631
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
632
				struct nfs4_ol_stateid, st_perstateowner);
633 634 635 636
		release_lock_stateid(stp);
	}
}

637
static void release_lockowner(struct nfs4_lockowner *lo)
638
{
639 640
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
641 642 643
}

static void
644
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
645
{
646
	struct nfs4_lockowner *lo;
647 648

	while (!list_empty(&open_stp->st_lockowners)) {
649 650 651
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
652 653 654
	}
}

655
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
656 657 658
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
659 660 661 662 663 664
	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
665
	unhash_stid(&stp->st_stid);
666 667 668
	free_generic_stateid(stp);
}

669
static void unhash_openowner(struct nfs4_openowner *oo)
670
{
671
	struct nfs4_ol_stateid *stp;
672

673 674 675 676
	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,
677
				struct nfs4_ol_stateid, st_perstateowner);
678
		release_open_stateid(stp);
679 680 681
	}
}

682 683 684 685 686
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
687
		unhash_stid(&s->st_stid);
688 689 690 691 692
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

693
static void release_openowner(struct nfs4_openowner *oo)
694
{
695 696
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
697
	release_last_closed_stateid(oo);
698
	nfs4_free_openowner(oo);
699 700
}

M
Marc Eshel 已提交
701 702 703 704 705 706 707 708
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

709
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
710 711 712 713 714 715
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]);
}
716 717 718 719 720 721 722
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
/*
 * Bump the seqid on cstate->replay_owner, and clear replay_owner if it
 * won't be used for replay.
 */
void nfsd4_bump_seqid(struct nfsd4_compound_state *cstate, __be32 nfserr)
{
	struct nfs4_stateowner *so = cstate->replay_owner;

	if (nfserr == nfserr_replay_me)
		return;

	if (!seqid_mutating_err(ntohl(nfserr))) {
		cstate->replay_owner = NULL;
		return;
	}
	if (!so)
		return;
	if (so->so_is_open_owner)
		release_last_closed_stateid(openowner(so));
	so->so_seqid++;
	return;
}
M
Marc Eshel 已提交
745

A
Andy Adamson 已提交
746 747 748 749 750 751 752 753 754 755 756 757 758
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;
}

/*
759 760 761 762 763 764 765 766 767 768 769 770 771
 * 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)

772 773 774 775 776 777 778 779 780
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]);
}

781
/*
782 783 784 785 786 787 788 789
 * 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;
}

790
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
791
{
792 793
	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
794

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

798 799
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
800
 * re-negotiate active sessions and reduce their slot usage to make
801
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
802
 */
803
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
804
{
805
	int avail;
A
Andy Adamson 已提交
806

807
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
808

809
	spin_lock(&nfsd_drc_lock);
810 811
	avail = min((unsigned long)NFSD_MAX_MEM_PER_SESSION,
		    nfsd_drc_max_mem - nfsd_drc_mem_used);
812 813 814
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
815

816 817
	return num;
}
A
Andy Adamson 已提交
818

819 820
static void nfsd4_put_drc_mem(int slotsize, int num)
{
821
	spin_lock(&nfsd_drc_lock);
822
	nfsd_drc_mem_used -= slotsize * num;
823
	spin_unlock(&nfsd_drc_lock);
824
}
A
Andy Adamson 已提交
825

826
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
827 828 829
{
	struct nfsd4_session *new;
	int mem, i;
830

831 832 833
	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
834

835 836 837
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
838
	/* allocate each struct nfsd4_slot and data cache in one piece */
839 840 841 842
	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])
843 844
			goto out_free;
	}
845 846 847 848 849 850
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
851 852
}

853 854 855 856
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
857
{
858
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
859

860
	new->maxreqs = numslots;
861 862
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
863 864 865 866
	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
867

868 869 870 871 872
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
873

874 875 876 877
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
878

879 880 881 882 883
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
884
	nfsd4_probe_callback(clp);
885
	spin_unlock(&clp->cl_lock);
886
}
A
Andy Adamson 已提交
887

888
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
889 890
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
891

892 893
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
894
		return NULL;
895 896
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
897
	conn->cn_flags = flags;
898 899 900
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
901

902 903 904 905
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
906 907
}

908
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
909
{
910
	struct nfs4_client *clp = ses->se_client;
911

912
	spin_lock(&clp->cl_lock);
913
	__nfsd4_hash_conn(conn, ses);
914
	spin_unlock(&clp->cl_lock);
915 916
}

917
static int nfsd4_register_conn(struct nfsd4_conn *conn)
918
{
919
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
920
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
921 922
}

923
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
924
{
925
	int ret;
A
Andy Adamson 已提交
926

927
	nfsd4_hash_conn(conn, ses);
928 929 930 931
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
932
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
933 934 935
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
936
}
A
Andy Adamson 已提交
937

938
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
939 940 941
{
	u32 dir = NFS4_CDFC4_FORE;

942
	if (cses->flags & SESSION4_BACK_CHAN)
943
		dir |= NFS4_CDFC4_BACK;
944
	return alloc_conn(rqstp, dir);
945 946 947
}

/* must be called under client_lock */
948
static void nfsd4_del_conns(struct nfsd4_session *s)
949
{
950 951
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
952

953 954 955 956 957
	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);
958

959 960
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
961

962 963 964
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
965
}
A
Andy Adamson 已提交
966

967 968 969 970 971 972 973
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);
}

974
static void free_session(struct nfsd4_session *ses)
975
{
976
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);
977 978

	lockdep_assert_held(&nn->client_lock);
979
	nfsd4_del_conns(ses);
980
	__free_session(ses);
981 982
}

983 984
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan,
					   struct nfsd_net *nn)
985 986 987 988 989 990 991 992 993 994 995 996
{
	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);
997 998
	if (numslots < 1)
		return NULL;
999

1000
	new = __alloc_session(slotsize, numslots);
1001
	if (!new) {
1002
		nfsd4_put_drc_mem(slotsize, numslots);
1003
		return NULL;
1004
	}
1005
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize, nn);
1006 1007
	return new;
}
1008

1009
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
1010 1011
{
	int idx;
1012
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1013

A
Andy Adamson 已提交
1014 1015 1016
	new->se_client = clp;
	gen_sessionid(new);

1017 1018
	INIT_LIST_HEAD(&new->se_conns);

1019
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
1020
	new->se_flags = cses->flags;
1021
	new->se_cb_prog = cses->callback_prog;
1022
	new->se_cb_sec = cses->cb_sec;
1023
	atomic_set(&new->se_ref, 0);
1024
	idx = hash_sessionid(&new->se_sessionid);
1025
	spin_lock(&nn->client_lock);
1026
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
1027
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
1028
	list_add(&new->se_perclnt, &clp->cl_sessions);
1029
	spin_unlock(&clp->cl_lock);
1030
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
1031

1032
	if (cses->flags & SESSION4_BACK_CHAN) {
1033
		struct sockaddr *sa = svc_addr(rqstp);
1034 1035 1036 1037 1038 1039 1040
		/*
		 * 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:
		 */
1041 1042 1043
		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
1044 1045
}

1046
/* caller must hold client_lock */
M
Marc Eshel 已提交
1047
static struct nfsd4_session *
1048
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
1049 1050 1051
{
	struct nfsd4_session *elem;
	int idx;
1052
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
1053 1054 1055 1056

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
1057
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

1068
/* caller must hold client_lock */
A
Andy Adamson 已提交
1069
static void
M
Marc Eshel 已提交
1070
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
1071 1072
{
	list_del(&ses->se_hash);
1073
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
1074
	list_del(&ses->se_perclnt);
1075
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
1076 1077
}

L
Linus Torvalds 已提交
1078 1079
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1080
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1081
{
1082
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1083
		return 0;
A
Andy Adamson 已提交
1084
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1085
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1086 1087 1088 1089 1090 1091 1092 1093
	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.
 */
1094
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1095 1096 1097
{
	struct nfs4_client *clp;

1098 1099 1100
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1101
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1102 1103 1104
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1105
	}
1106
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1107 1108 1109 1110 1111 1112
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1113
	struct nfsd_net __maybe_unused *nn = net_generic(clp->net, nfsd_net_id);
1114 1115

	lockdep_assert_held(&nn->client_lock);
1116 1117 1118 1119 1120
	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);
1121 1122
		WARN_ON_ONCE(atomic_read(&ses->se_ref));
		free_session(ses);
1123
	}
1124
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1125
	kfree(clp->cl_name.data);
M
majianpeng 已提交
1126
	idr_destroy(&clp->cl_stateids);
L
Linus Torvalds 已提交
1127 1128 1129
	kfree(clp);
}

B
Benny Halevy 已提交
1130 1131 1132 1133
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1134 1135
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1136
	list_del(&clp->cl_lru);
1137
	spin_lock(&clp->cl_lock);
1138 1139
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1140
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1141 1142
}

L
Linus Torvalds 已提交
1143
static void
1144
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1145
{
1146
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1147 1148
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1149
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1150 1151 1152

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1153 1154 1155
	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
1156 1157 1158 1159 1160 1161 1162
		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);
	}
1163
	while (!list_empty(&clp->cl_openowners)) {
1164 1165
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1166
	}
1167
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1168 1169
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1170
	list_del(&clp->cl_idhash);
1171
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1172
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1173
	else
1174
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1175
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1176
	unhash_client_locked(clp);
1177 1178
	WARN_ON_ONCE(atomic_read(&clp->cl_refcount));
	free_client(clp);
1179
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1180 1181
}

1182 1183 1184 1185 1186 1187
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1188 1189 1190 1191
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
1192 1193
}

1194 1195
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1196 1197 1198 1199
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1200
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1201
{
1202 1203 1204 1205 1206 1207 1208
	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;
1209
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1210 1211 1212 1213
	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);
1214
	return 0;
L
Linus Torvalds 已提交
1215 1216
}

1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
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);
}

1228
static int same_name(const char *n1, const char *n2)
1229
{
N
NeilBrown 已提交
1230
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1231 1232 1233
}

static int
1234 1235 1236
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1237 1238 1239
}

static int
1240 1241 1242
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
1243 1244
}

1245 1246 1247 1248 1249 1250 1251
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++)
1252
		if (!gid_eq(GROUP_AT(g1, i), GROUP_AT(g2, i)))
1253 1254 1255 1256
			return false;
	return true;
}

1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
/*
 * 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);
}


1273
static bool
1274 1275
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1276
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1277 1278
		|| (!uid_eq(cr1->cr_uid, cr2->cr_uid))
		|| (!gid_eq(cr1->cr_gid, cr2->cr_gid))
1279 1280 1281 1282 1283 1284
		|| !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;
1285
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1286 1287
}

1288
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1289 1290 1291
{
	static u32 current_clientid = 1;

1292
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1293 1294 1295
	clp->cl_clientid.cl_id = current_clientid++; 
}

1296 1297
static void gen_confirm(struct nfs4_client *clp)
{
1298
	__be32 verf[2];
1299
	static u32 i;
L
Linus Torvalds 已提交
1300

1301 1302 1303
	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
1304 1305
}

1306
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1307
{
J
J. Bruce Fields 已提交
1308 1309 1310 1311 1312 1313
	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
1314 1315
}

1316
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1317 1318
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1319

1320
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1321 1322
	if (!s)
		return NULL;
1323
	if (typemask & s->sc_type)
1324 1325 1326 1327
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1328
static struct nfs4_client *create_client(struct xdr_netobj name,
1329 1330 1331 1332
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1333
	int ret;
1334
	struct net *net = SVC_NET(rqstp);
1335
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1336 1337 1338 1339 1340

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

1341
	INIT_LIST_HEAD(&clp->cl_sessions);
1342 1343
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1344
		spin_lock(&nn->client_lock);
1345
		free_client(clp);
1346
		spin_unlock(&nn->client_lock);
1347
		return NULL;
1348
	}
1349
	idr_init(&clp->cl_stateids);
1350
	atomic_set(&clp->cl_refcount, 0);
1351
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1352 1353 1354 1355
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1356
	INIT_LIST_HEAD(&clp->cl_callbacks);
1357
	spin_lock_init(&clp->cl_lock);
1358
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1359
	clp->cl_time = get_seconds();
1360 1361 1362 1363 1364
	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);
1365
	clp->cl_cb_session = NULL;
1366
	clp->net = net;
1367 1368 1369
	return clp;
}

1370
static void
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
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
1412 1413
{
	unsigned int idhashval;
1414
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1415

1416
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1417
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1418
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1419
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1420
	renew_client(clp);
L
Linus Torvalds 已提交
1421 1422
}

1423
static void
L
Linus Torvalds 已提交
1424 1425 1426
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1427
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1428 1429

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1430
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1431
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1432
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1433
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1434 1435 1436 1437
	renew_client(clp);
}

static struct nfs4_client *
J
J. Bruce Fields 已提交
1438
find_client_in_id_table(struct list_head *tbl, clientid_t *clid, bool sessions)
L
Linus Torvalds 已提交
1439 1440 1441 1442
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

J
J. Bruce Fields 已提交
1443
	list_for_each_entry(clp, &tbl[idhashval], cl_idhash) {
1444
		if (same_clid(&clp->cl_clientid, clid)) {
1445 1446
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1447
			renew_client(clp);
L
Linus Torvalds 已提交
1448
			return clp;
1449
		}
L
Linus Torvalds 已提交
1450 1451 1452 1453
	}
	return NULL;
}

J
J. Bruce Fields 已提交
1454 1455 1456 1457 1458 1459 1460 1461
static struct nfs4_client *
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
{
	struct list_head *tbl = nn->conf_id_hashtbl;

	return find_client_in_id_table(tbl, clid, sessions);
}

L
Linus Torvalds 已提交
1462
static struct nfs4_client *
1463
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1464
{
J
J. Bruce Fields 已提交
1465
	struct list_head *tbl = nn->unconf_id_hashtbl;
L
Linus Torvalds 已提交
1466

J
J. Bruce Fields 已提交
1467
	return find_client_in_id_table(tbl, clid, sessions);
L
Linus Torvalds 已提交
1468 1469
}

1470
static bool clp_used_exchangeid(struct nfs4_client *clp)
1471
{
1472
	return clp->cl_exchange_flags != 0;
1473
} 
1474

1475
static struct nfs4_client *
1476
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1477
{
1478
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1479 1480 1481
}

static struct nfs4_client *
1482
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1483
{
1484
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1485 1486
}

1487
static void
1488
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1489
{
1490
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1491 1492
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
	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
1503 1504
		goto out_err;

1505
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1506
					    se->se_callback_addr_len,
1507 1508
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1509

1510
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1511
		goto out_err;
1512

1513 1514
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1515

1516 1517
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1518
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1519 1520
	return;
out_err:
1521 1522
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1523
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1524 1525 1526 1527 1528 1529
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1530
/*
1531
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1532 1533 1534 1535
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1536 1537
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1538

1539
	dprintk("--> %s slot %p\n", __func__, slot);
1540

1541 1542
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1543

1544
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1545
	if (nfsd4_not_cached(resp)) {
1546
		slot->sl_datalen = 0;
1547
		return;
1548
	}
1549 1550 1551 1552 1553 1554 1555
	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;
1556 1557 1558
}

/*
1559 1560 1561 1562
 * 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.
1563 1564
 *
 */
1565 1566 1567
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1568
{
1569 1570
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1571

1572 1573 1574
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1575

1576
	/* Return nfserr_retry_uncached_rep in next operation. */
1577
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1578 1579 1580
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1581
	}
1582
	return op->status;
1583 1584 1585
}

/*
1586 1587
 * The sequence operation is not cached because we can use the slot and
 * session values.
1588 1589
 */
__be32
1590 1591
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1592
{
1593
	struct nfsd4_slot *slot = resp->cstate.slot;
1594 1595
	__be32 status;

1596
	dprintk("--> %s slot %p\n", __func__, slot);
1597

1598 1599 1600 1601
	/* 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;
1602

1603 1604
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1605

1606 1607 1608
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1609 1610 1611 1612

	return status;
}

A
Andy Adamson 已提交
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628
/*
 * 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;
}

1629 1630 1631 1632 1633 1634 1635 1636
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.
	 */
1637 1638 1639
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1640 1641
}

A
Andy Adamson 已提交
1642 1643 1644 1645 1646
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1647
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1648
	__be32 status;
1649
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1650
	nfs4_verifier		verf = exid->verifier;
1651
	struct sockaddr		*sa = svc_addr(rqstp);
1652
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1653
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1654

1655
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1656
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1657
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1658
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1659
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1660

1661
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1662 1663 1664 1665 1666 1667
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1668 1669
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1670 1671 1672 1673 1674
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1675
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1676
	nfs4_lock_state();
1677
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1678
	if (conf) {
1679 1680 1681
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1682 1683
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1684
				status = nfserr_inval;
1685 1686
				goto out;
			}
1687
			if (!creds_match) { /* case 9 */
1688
				status = nfserr_perm;
1689 1690 1691
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1692 1693 1694
				status = nfserr_not_same;
				goto out;
			}
1695 1696 1697 1698
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1699
		}
1700
		if (!creds_match) { /* case 3 */
1701 1702
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1703 1704 1705 1706 1707
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1708
		if (verfs_match) { /* case 2 */
1709
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1710 1711 1712 1713 1714
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1715 1716
	}

1717
	if (update) { /* case 7 */
1718 1719
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1720 1721
	}

1722
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1723
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1724 1725
		expire_client(unconf);

1726
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1727
out_new:
J
Jeff Layton 已提交
1728
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1729
	if (new == NULL) {
1730
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1731 1732
		goto out;
	}
1733
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1734

1735
	gen_clid(new, nn);
1736
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1737 1738 1739 1740
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1741
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1742 1743 1744
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1745
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1746 1747 1748 1749 1750
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1751 1752
}

J
J. Bruce Fields 已提交
1753
static __be32
1754
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1755
{
1756 1757
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1758 1759

	/* The slot is in use, and no response has been sent. */
1760 1761
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1762 1763 1764 1765
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1766
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1767
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1768
		return nfs_ok;
1769
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1770 1771 1772 1773
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1774 1775 1776 1777 1778 1779 1780
/*
 * 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
1781
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794
{
	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;
}

1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
#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
1812
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1813 1814 1815 1816 1817
{
	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
1818 1819 1820 1821 1822
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1823
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1824
	struct nfs4_client *conf, *unconf;
1825
	struct nfsd4_session *new;
1826
	struct nfsd4_conn *conn;
1827
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1828
	__be32 status = 0;
1829
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1830

1831 1832
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1833 1834
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1835
	new = alloc_session(&cr_ses->fore_channel, nn);
1836 1837 1838 1839 1840 1841
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1842

A
Andy Adamson 已提交
1843
	nfs4_lock_state();
1844
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1845
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
1846
	WARN_ON_ONCE(conf && unconf);
A
Andy Adamson 已提交
1847 1848

	if (conf) {
1849 1850
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1851
		if (status == nfserr_replay_cache) {
1852
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1853
			goto out_free_conn;
1854
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1855
			status = nfserr_seq_misordered;
1856
			goto out_free_conn;
A
Andy Adamson 已提交
1857 1858
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1859
		struct nfs4_client *old;
A
Andy Adamson 已提交
1860
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1861
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1862
			status = nfserr_clid_inuse;
1863
			goto out_free_conn;
A
Andy Adamson 已提交
1864
		}
1865 1866
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1867 1868
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1869
			status = nfserr_seq_misordered;
1870
			goto out_free_conn;
A
Andy Adamson 已提交
1871
		}
1872
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
1873 1874 1875 1876
		if (old) {
			status = mark_client_expired(old);
			if (status)
				goto out_free_conn;
J
J. Bruce Fields 已提交
1877
			expire_client(old);
1878
		}
J
J. Bruce Fields 已提交
1879
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1880 1881 1882
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1883
		goto out_free_conn;
A
Andy Adamson 已提交
1884
	}
1885
	status = nfs_ok;
1886 1887 1888 1889 1890 1891
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1892 1893 1894
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1895
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1896
	       NFS4_MAX_SESSIONID_LEN);
1897 1898
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1899
	cs_slot->sl_seqid++;
1900
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1901

1902 1903
	/* 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
1904 1905
	nfs4_unlock_state();
	return status;
1906
out_free_conn:
1907
	nfs4_unlock_state();
1908 1909 1910
	free_conn(conn);
out_free_session:
	__free_session(new);
J
J. Bruce Fields 已提交
1911
	return status;
A
Andy Adamson 已提交
1912 1913
}

1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927
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;
}

1928 1929 1930
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1931
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1932

1933
	spin_lock(&nn->client_lock);
1934 1935
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1936
	spin_unlock(&nn->client_lock);
1937 1938 1939 1940 1941 1942

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1943 1944 1945 1946 1947
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1948
	struct nfsd4_conn *conn;
1949
	struct nfsd4_session *session;
1950
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1951 1952 1953

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1954
	nfs4_lock_state();
1955
	spin_lock(&nn->client_lock);
1956
	session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1957
	spin_unlock(&nn->client_lock);
1958 1959 1960
	status = nfserr_badsession;
	if (!session)
		goto out;
1961
	status = nfsd4_map_bcts_dir(&bcts->dir);
1962
	if (status)
1963
		goto out;
1964
	conn = alloc_conn(rqstp, bcts->dir);
1965
	status = nfserr_jukebox;
1966
	if (!conn)
1967 1968 1969 1970 1971 1972
		goto out;
	nfsd4_init_conn(rqstp, conn, session);
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
1973 1974
}

1975 1976 1977 1978 1979 1980 1981
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
1982 1983 1984 1985 1986
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1987
	struct nfsd4_session *ses;
1988
	__be32 status;
1989
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1990

1991 1992
	nfs4_lock_state();
	status = nfserr_not_only_op;
1993
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1994
		if (!nfsd4_last_compound_op(r))
1995
			goto out;
1996
	}
B
Benny Halevy 已提交
1997
	dump_sessionid(__func__, &sessionid->sessionid);
1998
	spin_lock(&nn->client_lock);
1999
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
2000 2001 2002
	status = nfserr_badsession;
	if (!ses)
		goto out_client_lock;
2003 2004 2005
	status = mark_session_dead_locked(ses);
	if (status)
		goto out_client_lock;
B
Benny Halevy 已提交
2006
	unhash_session(ses);
2007
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2008

2009
	nfsd4_probe_callback_sync(ses->se_client);
2010

2011
	spin_lock(&nn->client_lock);
2012
	free_session(ses);
B
Benny Halevy 已提交
2013
	status = nfs_ok;
2014 2015
out_client_lock:
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2016
out:
2017
	nfs4_unlock_state();
B
Benny Halevy 已提交
2018
	return status;
A
Andy Adamson 已提交
2019 2020
}

2021
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
2022 2023 2024 2025
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
2026
		if (c->cn_xprt == xpt) {
2027 2028 2029 2030 2031 2032
			return c;
		}
	}
	return NULL;
}

2033
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
2034 2035
{
	struct nfs4_client *clp = ses->se_client;
2036
	struct nfsd4_conn *c;
2037
	int ret;
2038 2039

	spin_lock(&clp->cl_lock);
2040
	c = __nfsd4_find_conn(new->cn_xprt, ses);
2041 2042 2043 2044 2045 2046 2047
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
2048 2049 2050 2051
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2052 2053 2054
	return;
}

2055 2056 2057 2058 2059 2060 2061
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
2062 2063 2064 2065 2066 2067 2068 2069
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
2070
__be32
B
Benny Halevy 已提交
2071
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2072 2073 2074
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2075
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2076
	struct nfsd4_session *session;
2077
	struct nfs4_client *clp;
B
Benny Halevy 已提交
2078
	struct nfsd4_slot *slot;
2079
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2080
	__be32 status;
2081
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2082

2083 2084 2085
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2086 2087 2088 2089 2090 2091 2092 2093
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2094
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2095
	status = nfserr_badsession;
2096
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2097
	if (!session)
2098 2099 2100 2101 2102
		goto out_no_session;
	clp = session->se_client;
	status = get_client_locked(clp);
	if (status)
		goto out_no_session;
2103 2104 2105
	status = nfsd4_get_session_locked(session);
	if (status)
		goto out_put_client;
B
Benny Halevy 已提交
2106

2107 2108
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
2109
		goto out_put_session;
2110

M
Mi Jinlong 已提交
2111 2112
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
2113
		goto out_put_session;
M
Mi Jinlong 已提交
2114

B
Benny Halevy 已提交
2115
	status = nfserr_badslot;
2116
	if (seq->slotid >= session->se_fchannel.maxreqs)
2117
		goto out_put_session;
B
Benny Halevy 已提交
2118

2119
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2120 2121
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2122 2123 2124 2125 2126
	/* 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;

2127 2128
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2129
	if (status == nfserr_replay_cache) {
2130 2131
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
2132
			goto out_put_session;
B
Benny Halevy 已提交
2133 2134
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2135
		/* Return the cached reply status and set cstate->status
2136
		 * for nfsd4_proc_compound processing */
2137
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2138
		cstate->status = nfserr_replay_cache;
2139
		goto out;
B
Benny Halevy 已提交
2140 2141
	}
	if (status)
2142
		goto out_put_session;
B
Benny Halevy 已提交
2143

2144 2145
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2146

B
Benny Halevy 已提交
2147 2148
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2149
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2150 2151
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2152 2153
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2154 2155 2156 2157 2158

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

out:
2159 2160 2161 2162 2163 2164 2165 2166 2167
	switch (clp->cl_cb_state) {
	case NFSD4_CB_DOWN:
		seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
		break;
	case NFSD4_CB_FAULT:
		seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
		break;
	default:
		seq->status_flags = 0;
2168
	}
2169
out_no_session:
2170
	kfree(conn);
2171
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2172
	return status;
2173 2174
out_put_session:
	nfsd4_put_session(session);
2175 2176 2177
out_put_client:
	put_client_renew_locked(clp);
	goto out_no_session;
A
Andy Adamson 已提交
2178 2179
}

2180 2181 2182 2183
__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
2184
	__be32 status = 0;
2185
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2186 2187

	nfs4_lock_state();
2188
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2189
	conf = find_confirmed_client(&dc->clientid, true, nn);
2190
	WARN_ON_ONCE(conf && unconf);
2191 2192 2193 2194

	if (conf) {
		clp = conf;

2195
		if (client_has_state(conf)) {
2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211
			status = nfserr_clientid_busy;
			goto out;
		}
	} else if (unconf)
		clp = unconf;
	else {
		status = nfserr_stale_clientid;
		goto out;
	}

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

2212 2213 2214
__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
2215
	__be32 status = 0;
2216

2217 2218 2219 2220 2221 2222 2223 2224 2225
	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;
	}
2226

2227
	nfs4_lock_state();
2228
	status = nfserr_complete_already;
2229 2230
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2231 2232 2233 2234
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2235 2236 2237 2238 2239 2240 2241
		/*
		 * 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.
		 */
2242 2243 2244
		goto out;

	status = nfs_ok;
2245
	nfsd4_client_record_create(cstate->session->se_client);
2246
out:
2247
	nfs4_unlock_state();
2248
	return status;
2249 2250
}

2251
__be32
2252 2253
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2254
{
2255
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2256
	nfs4_verifier		clverifier = setclid->se_verf;
2257
	struct nfs4_client	*conf, *unconf, *new;
2258
	__be32 			status;
2259 2260
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2261
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2262
	nfs4_lock_state();
2263
	conf = find_confirmed_client_by_name(&clname, nn);
2264
	if (conf) {
2265
		/* case 0: */
L
Linus Torvalds 已提交
2266
		status = nfserr_clid_inuse;
2267 2268
		if (clp_used_exchangeid(conf))
			goto out;
2269
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2270 2271 2272 2273 2274
			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
2275 2276 2277
			goto out;
		}
	}
2278
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2279 2280
	if (unconf)
		expire_client(unconf);
2281
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2282
	new = create_client(clname, rqstp, &clverifier);
2283 2284
	if (new == NULL)
		goto out;
2285
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2286
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2287
		copy_clid(new, conf);
2288
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2289
		gen_clid(new, nn);
2290
	new->cl_minorversion = 0;
2291
	gen_callback(new, setclid, rqstp);
2292
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2293 2294 2295 2296 2297 2298 2299 2300 2301 2302
	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;
}


2303
__be32
2304 2305 2306
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2307
{
2308
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2309 2310
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2311
	__be32 status;
2312
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2313

2314
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2315 2316
		return nfserr_stale_clientid;
	nfs4_lock_state();
2317

2318
	conf = find_confirmed_client(clid, false, nn);
2319
	unconf = find_unconfirmed_client(clid, false, nn);
2320
	/*
2321 2322 2323 2324
	 * 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.
2325
	 */
2326 2327 2328 2329 2330
	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;
2331
	/* cases below refer to rfc 3530 section 14.2.34: */
2332 2333
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2334
			status = nfs_ok;
2335 2336 2337 2338 2339 2340
		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 */
2341 2342 2343
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2344
	} else { /* case 3: normal case; new or rebooted client */
2345
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2346 2347 2348 2349
		if (conf) {
			status = mark_client_expired(conf);
			if (status)
				goto out;
2350
			expire_client(conf);
2351
		}
2352
		move_to_confirmed(unconf);
2353
		nfsd4_probe_callback(unconf);
2354
	}
L
Linus Torvalds 已提交
2355 2356 2357 2358 2359
out:
	nfs4_unlock_state();
	return status;
}

2360 2361 2362 2363 2364
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2365
/* OPEN Share state helper functions */
2366
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2367 2368 2369
{
	unsigned int hashval = file_hashval(ino);

2370 2371 2372 2373 2374 2375 2376 2377 2378
	atomic_set(&fp->fi_ref, 1);
	INIT_LIST_HEAD(&fp->fi_stateids);
	INIT_LIST_HEAD(&fp->fi_delegations);
	fp->fi_inode = igrab(ino);
	fp->fi_had_conflict = false;
	fp->fi_lease = NULL;
	memset(fp->fi_fds, 0, sizeof(fp->fi_fds));
	memset(fp->fi_access, 0, sizeof(fp->fi_access));
	spin_lock(&recall_lock);
2379
	hlist_add_head(&fp->fi_hash, &file_hashtbl[hashval]);
2380
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2381 2382
}

N
NeilBrown 已提交
2383
static void
2384
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2385
{
N
NeilBrown 已提交
2386 2387
	if (*slab == NULL)
		return;
2388
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2389
	*slab = NULL;
L
Linus Torvalds 已提交
2390 2391
}

2392
void
L
Linus Torvalds 已提交
2393 2394
nfsd4_free_slabs(void)
{
2395 2396
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2397
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2398
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2399
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2400
}
L
Linus Torvalds 已提交
2401

2402
int
N
NeilBrown 已提交
2403 2404
nfsd4_init_slabs(void)
{
2405 2406 2407 2408 2409
	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",
2410
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2411
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2412 2413
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2414
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2415 2416
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2417
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2418
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2419 2420
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2421
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2422
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2423 2424
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2425 2426 2427 2428 2429
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2430 2431
}

2432 2433 2434 2435 2436 2437 2438
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
2439
{
2440 2441
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2442 2443
}

2444
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2445
{
2446 2447 2448
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2449 2450
}

2451
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2452
{
L
Linus Torvalds 已提交
2453 2454
	struct nfs4_stateowner *sop;

2455
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2456 2457 2458 2459 2460
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2461
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2462
		return NULL;
2463 2464 2465
	}
	sop->so_owner.len = owner->len;

2466
	INIT_LIST_HEAD(&sop->so_stateids);
2467 2468 2469 2470 2471
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2472
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2473
{
2474 2475 2476
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2477
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2478 2479
}

2480
static struct nfs4_openowner *
2481
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2482
	struct nfs4_openowner *oo;
2483

2484 2485
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2486
		return NULL;
2487 2488
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2489
	oo->oo_flags = NFS4_OO_NEW;
2490
	oo->oo_time = 0;
2491
	oo->oo_last_closed_stid = NULL;
2492 2493 2494
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2495 2496
}

2497
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2498
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2499

J
J. Bruce Fields 已提交
2500
	stp->st_stid.sc_type = NFS4_OPEN_STID;
2501
	INIT_LIST_HEAD(&stp->st_lockowners);
2502
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2503
	list_add(&stp->st_perfile, &fp->fi_stateids);
2504
	stp->st_stateowner = &oo->oo_owner;
2505
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2506 2507 2508
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2509
	set_access(open->op_share_access, stp);
2510
	set_deny(open->op_share_deny, stp);
2511
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2512 2513
}

2514
static void
2515
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2516
{
2517 2518
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2519
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2520

2521
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2522
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2523 2524 2525
}

static int
2526 2527 2528 2529 2530 2531
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
2532 2533
}

2534
static struct nfs4_openowner *
2535 2536
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2537
{
2538 2539
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2540
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2541

2542
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2543 2544
		if (!so->so_is_open_owner)
			continue;
2545 2546
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2547 2548 2549
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2550 2551 2552
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563
	}
	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;

2564
	spin_lock(&recall_lock);
2565
	hlist_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2566 2567
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2568
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2569
			return fp;
2570
		}
L
Linus Torvalds 已提交
2571
	}
2572
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2573 2574 2575 2576 2577 2578 2579
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2580
static __be32
L
Linus Torvalds 已提交
2581 2582 2583 2584
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;
2585
	struct nfs4_ol_stateid *stp;
2586
	__be32 ret;
L
Linus Torvalds 已提交
2587 2588

	fp = find_file(ino);
2589 2590
	if (!fp)
		return nfs_ok;
2591
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2592
	/* Search for conflicting share reservations */
2593
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2594 2595
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2596
			goto out;
L
Linus Torvalds 已提交
2597
	}
2598 2599 2600 2601
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2602 2603
}

2604
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2605
{
2606 2607 2608
	struct nfs4_client *clp = dp->dl_stid.sc_client;
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

L
Linus Torvalds 已提交
2609 2610 2611 2612 2613 2614 2615
	/* 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);

2616
	list_add_tail(&dp->dl_recall_lru, &nn->del_recall_lru);
L
Linus Torvalds 已提交
2617

2618
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2619 2620
	dp->dl_time = get_seconds();

2621 2622 2623
	nfsd4_cb_recall(dp);
}

2624
/* Called from break_lease() with lock_flocks() held. */
2625 2626
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2627 2628
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2629

2630 2631 2632 2633 2634 2635 2636 2637
	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;
	}
2638 2639
	/*
	 * We don't want the locks code to timeout the lease for us;
2640
	 * we'll remove it ourself if a delegation isn't returned
2641
	 * in time:
2642 2643
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2644

2645
	spin_lock(&recall_lock);
2646 2647 2648
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2649
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660
}

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

2661
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2662 2663
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2664 2665
};

2666 2667 2668 2669 2670 2671 2672 2673 2674 2675
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
2676

2677
__be32
A
Andy Adamson 已提交
2678
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2679
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2680 2681 2682 2683
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2684
	struct nfs4_openowner *oo = NULL;
2685
	__be32 status;
L
Linus Torvalds 已提交
2686

2687
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2688
		return nfserr_stale_clientid;
2689 2690 2691 2692 2693 2694 2695
	/*
	 * 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
2696

2697
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2698
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2699 2700
	open->op_openowner = oo;
	if (!oo) {
2701 2702
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2703
		if (clp == NULL)
2704
			return nfserr_expired;
2705
		goto new_owner;
L
Linus Torvalds 已提交
2706
	}
2707
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2708
		/* Replace unconfirmed owners without checking for replay. */
2709 2710 2711
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2712
		goto new_owner;
2713
	}
2714 2715 2716 2717 2718
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2719 2720 2721 2722 2723
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2724 2725 2726 2727
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2728
	return nfs_ok;
L
Linus Torvalds 已提交
2729 2730
}

2731
static inline __be32
N
NeilBrown 已提交
2732 2733 2734 2735 2736 2737 2738 2739
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;
}

2740
static int share_access_to_flags(u32 share_access)
2741
{
2742
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2743 2744
}

2745
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2746
{
2747
	struct nfs4_stid *ret;
2748

2749
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2750 2751 2752
	if (!ret)
		return NULL;
	return delegstateid(ret);
2753 2754
}

2755 2756 2757 2758 2759 2760
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;
}

2761
static __be32
2762
nfs4_check_deleg(struct nfs4_client *cl, struct nfsd4_open *open,
2763 2764 2765
		struct nfs4_delegation **dp)
{
	int flags;
2766
	__be32 status = nfserr_bad_stateid;
2767

2768
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2769
	if (*dp == NULL)
2770
		goto out;
2771
	flags = share_access_to_flags(open->op_share_access);
2772 2773 2774
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2775
out:
2776
	if (!nfsd4_is_deleg_cur(open))
2777 2778 2779
		return nfs_ok;
	if (status)
		return status;
2780
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2781
	return nfs_ok;
2782 2783
}

2784
static __be32
2785
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2786
{
2787
	struct nfs4_ol_stateid *local;
2788
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2789

2790
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2791 2792 2793 2794
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2795
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2796 2797 2798
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2799
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2800
	}
2801
	return nfs_ok;
L
Linus Torvalds 已提交
2802 2803
}

2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814
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;
}

2815 2816
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2817 2818
{
	__be32 status;
2819 2820 2821
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832
	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;
}

2833
static inline __be32
L
Linus Torvalds 已提交
2834 2835 2836 2837 2838 2839 2840 2841 2842 2843
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))
2844
		return nfserr_inval;
L
Linus Torvalds 已提交
2845 2846 2847
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2848
static __be32
2849
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
2850
{
2851
	u32 op_share_access = open->op_share_access;
2852
	bool new_access;
2853
	__be32 status;
L
Linus Torvalds 已提交
2854

2855
	new_access = !test_access(op_share_access, stp);
2856
	if (new_access) {
2857
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2858 2859
		if (status)
			return status;
2860
	}
L
Linus Torvalds 已提交
2861 2862
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2863
		if (new_access) {
2864
			int oflag = nfs4_access_to_omode(op_share_access);
2865 2866
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2867 2868 2869
		return status;
	}
	/* remember the open */
2870
	set_access(op_share_access, stp);
2871
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2872 2873 2874 2875 2876 2877

	return nfs_ok;
}


static void
2878
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2879
{
2880
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2881 2882
}

2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895
/* 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;
}

2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907
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;
2908
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2909 2910 2911 2912
	fl->fl_pid = current->tgid;
	return fl;
}

2913 2914
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2915
	struct nfs4_file *fp = dp->dl_file;
2916 2917 2918 2919 2920 2921
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2922
	fl->fl_file = find_readable_file(fp);
2923
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2924
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2925
	if (status) {
2926
		list_del_init(&dp->dl_perclnt);
2927 2928 2929
		locks_free_lock(fl);
		return -ENOMEM;
	}
2930
	fp->fi_lease = fl;
A
Al Viro 已提交
2931
	fp->fi_deleg_file = get_file(fl->fl_file);
2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950
	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);
2951
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2952 2953 2954
	return 0;
}

2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970
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:
2971
			WARN_ON_ONCE(1);
2972 2973 2974 2975
		}
	}
}

L
Linus Torvalds 已提交
2976 2977 2978 2979
/*
 * Attempt to hand out a delegation.
 */
static void
2980 2981
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
2982 2983
{
	struct nfs4_delegation *dp;
2984
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2985
	int cb_up;
2986
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2987

2988
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2989
	flag = NFS4_OPEN_DELEGATE_NONE;
2990 2991 2992
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2993
			if (!cb_up)
2994 2995 2996 2997 2998 2999 3000 3001
				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.... */
3002
			if (locks_in_grace(net))
3003
				goto out;
3004
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
3005 3006 3007 3008 3009 3010 3011 3012 3013
				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
3014

3015
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
3016 3017
	if (dp == NULL)
		goto out_no_deleg;
3018
	status = nfs4_set_delegation(dp, flag);
3019
	if (status)
3020
		goto out_free;
L
Linus Torvalds 已提交
3021

3022
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
3023

3024
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
3025
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
3026 3027
out:
	open->op_delegate_type = flag;
3028 3029 3030 3031 3032
	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");

3033 3034 3035
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
3036
	}
3037 3038
	return;
out_free:
3039
	unhash_stid(&dp->dl_stid);
3040
	nfs4_put_delegation(dp);
3041 3042 3043
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
3044 3045
}

3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063
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
3064 3065 3066
/*
 * called with nfs4_lock_state() held.
 */
3067
__be32
L
Linus Torvalds 已提交
3068 3069
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3070
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3071
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3072 3073
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3074
	struct nfs4_ol_stateid *stp = NULL;
3075
	struct nfs4_delegation *dp = NULL;
3076
	__be32 status;
L
Linus Torvalds 已提交
3077 3078 3079 3080 3081 3082 3083 3084 3085 3086

	/*
	 * 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;
3087
		status = nfs4_check_deleg(cl, open, &dp);
3088 3089
		if (status)
			goto out;
L
Linus Torvalds 已提交
3090
	} else {
3091
		status = nfserr_bad_stateid;
3092
		if (nfsd4_is_deleg_cur(open))
3093
			goto out;
3094
		status = nfserr_jukebox;
3095 3096 3097
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3098 3099 3100 3101 3102 3103 3104 3105
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3106
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3107 3108 3109
		if (status)
			goto out;
	} else {
3110
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3111 3112 3113
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3114
		if (status)
L
Linus Torvalds 已提交
3115
			goto out;
3116 3117
		stp = open->op_stp;
		open->op_stp = NULL;
3118
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3119
	}
3120 3121
	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
3122

3123
	if (nfsd4_has_session(&resp->cstate)) {
3124
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3125

3126 3127 3128 3129 3130 3131 3132
		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
3133 3134 3135 3136
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3137
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3138
nodeleg:
L
Linus Torvalds 已提交
3139 3140
	status = nfs_ok;

3141
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3142
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3143
out:
3144 3145
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3146 3147
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3148

3149 3150
	if (fp)
		put_nfs4_file(fp);
3151
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3152
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3153 3154 3155 3156
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3157
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3158
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3159 3160 3161 3162 3163
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178
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;
		}
	}
3179 3180
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3181
	if (open->op_stp)
3182
		free_generic_stateid(open->op_stp);
3183 3184
}

3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196
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;
}

3197
__be32
3198 3199
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3200 3201
{
	struct nfs4_client *clp;
3202
	__be32 status;
3203
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3204 3205 3206 3207

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3208 3209
	status = lookup_clientid(clid, cstate->minorversion, nn, &clp);
	if (status)
L
Linus Torvalds 已提交
3210 3211
		goto out;
	status = nfserr_cb_path_down;
3212
	if (!list_empty(&clp->cl_delegations)
3213
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3214 3215 3216 3217 3218 3219 3220
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3221
static void
3222
nfsd4_end_grace(struct nfsd_net *nn)
3223
{
3224
	/* do nothing if grace period already ended */
3225
	if (nn->grace_ended)
3226 3227
		return;

3228
	dprintk("NFSD: end of grace period\n");
3229
	nn->grace_ended = true;
3230
	nfsd4_record_grace_done(nn, nn->boot_time);
3231
	locks_end_grace(&nn->nfsd4_manager);
3232 3233 3234 3235 3236
	/*
	 * 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:
	 */
3237
	nn->nfsd4_grace = nn->nfsd4_lease;
3238 3239
}

3240
static time_t
3241
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3242 3243
{
	struct nfs4_client *clp;
3244
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3245 3246
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3247 3248 3249
	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
3250 3251 3252 3253

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3254
	nfsd4_end_grace(nn);
3255
	INIT_LIST_HEAD(&reaplist);
3256
	spin_lock(&nn->client_lock);
3257
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3258 3259 3260 3261 3262 3263 3264
		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;
		}
3265
		if (mark_client_expired_locked(clp)) {
3266 3267 3268 3269
			dprintk("NFSD: client in use (clientid %08x)\n",
				clp->cl_clientid.cl_id);
			continue;
		}
3270
		list_move(&clp->cl_lru, &reaplist);
3271
	}
3272
	spin_unlock(&nn->client_lock);
3273 3274
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3275 3276 3277 3278 3279
		dprintk("NFSD: purging unused client (clientid %08x)\n",
			clp->cl_clientid.cl_id);
		expire_client(clp);
	}
	spin_lock(&recall_lock);
3280
	list_for_each_safe(pos, next, &nn->del_recall_lru) {
L
Linus Torvalds 已提交
3281
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
3282 3283
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296
		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);
	}
3297
	test_val = nn->nfsd4_lease;
3298
	list_for_each_safe(pos, next, &nn->close_lru) {
3299 3300 3301
		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
3302 3303 3304 3305
			if (test_val > u)
				test_val = u;
			break;
		}
3306
		release_openowner(oo);
L
Linus Torvalds 已提交
3307 3308 3309 3310 3311 3312 3313
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3314 3315 3316 3317
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3318
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3319 3320
{
	time_t t;
3321 3322 3323 3324
	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
3325

3326
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3327
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3328
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3329 3330
}

3331
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3332
{
3333 3334 3335
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3336 3337 3338
}

static inline int
3339
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3340
{
3341 3342 3343
	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
3344 3345 3346
}

static inline int
3347
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3348
{
3349 3350
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3351 3352 3353
}

static
3354
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3355
{
3356
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3357

3358 3359 3360
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3361
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3362
                goto out;
3363
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3364 3365 3366 3367 3368 3369
                goto out;
	status = nfs_ok;
out:
	return status;
}

3370
static inline __be32
3371
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3372
{
3373
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3374
		return nfs_ok;
3375
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3376
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391
		 * 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
3392
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3393
{
3394
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3395 3396
}

3397 3398 3399 3400 3401 3402
/* 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
3403
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3404
{
A
Andy Adamson 已提交
3405 3406 3407 3408
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3409
	if (has_session && in->si_generation == 0)
3410 3411 3412 3413
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3414

3415
	/* If the client sends us a stateid from the future, it's buggy: */
3416
	if (stateid_generation_after(in, ref))
3417 3418
		return nfserr_bad_stateid;
	/*
3419 3420 3421 3422 3423 3424 3425 3426
	 * 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:
3427
	 */
3428
	return nfserr_old_stateid;
3429 3430
}

3431
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3432
{
3433 3434 3435
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3436

3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447
	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;
	}
3448
	s = find_stateid(cl, stateid);
3449
	if (!s)
3450
		return nfserr_bad_stateid;
3451
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3452
	if (status)
3453 3454 3455 3456 3457
		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
3458
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3459 3460
		return nfserr_bad_stateid;
	return nfs_ok;
3461 3462
}

3463 3464 3465
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3466 3467
{
	struct nfs4_client *cl;
J
J. Bruce Fields 已提交
3468
	__be32 status;
3469 3470 3471

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
J
J. Bruce Fields 已提交
3472 3473 3474
	status = lookup_clientid(&stateid->si_opaque.so_clid, sessions,
							nn, &cl);
	if (status == nfserr_stale_clientid)
3475
		return nfserr_stale_stateid;
J
J. Bruce Fields 已提交
3476 3477
	if (status)
		return status;
3478 3479 3480 3481 3482 3483
	*s = find_stateid_by_type(cl, stateid, typemask);
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;
}

L
Linus Torvalds 已提交
3484 3485 3486
/*
* Checks for stateid operations
*/
3487
__be32
3488
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3489
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3490
{
3491
	struct nfs4_stid *s;
3492
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3493
	struct nfs4_delegation *dp = NULL;
3494
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3495
	struct inode *ino = current_fh->fh_dentry->d_inode;
3496
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3497
	__be32 status;
L
Linus Torvalds 已提交
3498 3499 3500 3501

	if (filpp)
		*filpp = NULL;

3502
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3503 3504 3505
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3506
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3507

3508 3509
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3510 3511
	if (status)
		return status;
3512 3513 3514
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3515 3516
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3517
		dp = delegstateid(s);
3518 3519 3520
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3521
		if (filpp) {
3522
			*filpp = dp->dl_file->fi_deleg_file;
3523 3524 3525 3526 3527
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3528
		}
3529 3530 3531
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3532
		stp = openlockstateid(s);
3533 3534
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3535
			goto out;
3536
		if (stp->st_stateowner->so_is_open_owner
3537
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3538
			goto out;
3539 3540
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3541
			goto out;
3542 3543 3544 3545 3546 3547
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3548 3549 3550
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3551 3552 3553 3554 3555 3556
	}
	status = nfs_ok;
out:
	return status;
}

3557
static __be32
3558
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3559
{
3560
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3561 3562 3563 3564 3565
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3566 3567 3568 3569 3570 3571 3572
/*
 * 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)
{
3573 3574 3575 3576 3577
	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)
3578 3579
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3580 3581
	nfs4_unlock_state();

3582 3583 3584
	return nfs_ok;
}

3585 3586 3587 3588 3589
__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
3590
	struct nfs4_stid *s;
3591
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3592
	__be32 ret = nfserr_bad_stateid;
3593 3594

	nfs4_lock_state();
3595
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3596
	if (!s)
3597
		goto out;
J
J. Bruce Fields 已提交
3598 3599
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3600 3601
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3602 3603 3604 3605 3606 3607 3608 3609 3610
	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;
3611 3612 3613
		break;
	default:
		ret = nfserr_bad_stateid;
3614 3615 3616 3617 3618 3619
	}
out:
	nfs4_unlock_state();
	return ret;
}

3620 3621 3622 3623 3624 3625
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
3626

3627
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3628 3629 3630 3631 3632 3633 3634 3635
{
	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;
3636 3637 3638 3639 3640 3641 3642 3643 3644 3645
	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);
3646 3647
}

L
Linus Torvalds 已提交
3648 3649 3650
/* 
 * Checks for sequence id mutating operations. 
 */
3651
static __be32
3652
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3653
			 stateid_t *stateid, char typemask,
3654 3655
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3656
{
3657
	__be32 status;
3658
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3659

3660 3661
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3662

L
Linus Torvalds 已提交
3663
	*stpp = NULL;
3664 3665
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3666 3667
	if (status)
		return status;
3668
	*stpp = openlockstateid(s);
3669
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3670

3671 3672
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3673

3674 3675
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)
3676 3677 3678
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3679

3680
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3681
						NFS4_OPEN_STID, stpp, nn);
3682 3683
	if (status)
		return status;
3684
	oo = openowner((*stpp)->st_stateowner);
3685
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3686 3687
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3688 3689
}

3690
__be32
3691
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3692
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3693
{
3694
	__be32 status;
3695
	struct nfs4_openowner *oo;
3696
	struct nfs4_ol_stateid *stp;
3697
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3698 3699

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3700 3701
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3702

3703
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3704 3705
	if (status)
		return status;
L
Linus Torvalds 已提交
3706 3707 3708

	nfs4_lock_state();

3709
	status = nfs4_preprocess_seqid_op(cstate,
3710
					oc->oc_seqid, &oc->oc_req_stateid,
3711
					NFS4_OPEN_STID, &stp, nn);
3712
	if (status)
3713
		goto out;
3714
	oo = openowner(stp->st_stateowner);
3715
	status = nfserr_bad_stateid;
3716
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3717
		goto out;
3718
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3719 3720
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3721
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3722
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3723

3724
	nfsd4_client_record_create(oo->oo_owner.so_client);
3725
	status = nfs_ok;
L
Linus Torvalds 已提交
3726
out:
3727
	nfsd4_bump_seqid(cstate, status);
3728 3729
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3730 3731 3732
	return status;
}

J
J. Bruce Fields 已提交
3733
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3734
{
3735
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3736 3737
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3738
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3739
}
3740

J
J. Bruce Fields 已提交
3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754
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:
3755
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3756 3757 3758 3759
	}
}

static void
3760
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3761 3762 3763 3764
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3765
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3766 3767 3768
	}
}

3769
__be32
3770 3771
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3772
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3773
{
3774
	__be32 status;
3775
	struct nfs4_ol_stateid *stp;
3776
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3777 3778

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3779 3780
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3781

3782
	/* We don't yet support WANT bits: */
3783 3784 3785
	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
3786 3787

	nfs4_lock_state();
3788
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3789
					&od->od_stateid, &stp, nn);
3790
	if (status)
L
Linus Torvalds 已提交
3791 3792
		goto out; 
	status = nfserr_inval;
3793 3794
	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
3795 3796 3797
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3798
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3799 3800 3801 3802
		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
3803
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3804

3805
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3806

3807 3808
	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
3809 3810
	status = nfs_ok;
out:
3811
	nfsd4_bump_seqid(cstate, status);
3812 3813
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3814 3815 3816
	return status;
}

3817 3818 3819 3820
static void nfsd4_close_open_stateid(struct nfs4_ol_stateid *s)
{
	unhash_open_stateid(s);
	s->st_stid.sc_type = NFS4_CLOSED_STID;
3821 3822
}

L
Linus Torvalds 已提交
3823 3824 3825
/*
 * nfs4_unlock_state() called after encode
 */
3826
__be32
3827
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3828
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3829
{
3830
	__be32 status;
3831
	struct nfs4_openowner *oo;
3832
	struct nfs4_ol_stateid *stp;
3833 3834
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3835 3836

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3837 3838
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3839 3840

	nfs4_lock_state();
3841 3842 3843
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3844
					&stp, nn);
3845
	nfsd4_bump_seqid(cstate, status);
3846
	if (status)
L
Linus Torvalds 已提交
3847
		goto out; 
3848
	oo = openowner(stp->st_stateowner);
3849 3850
	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
3851

3852
	nfsd4_close_open_stateid(stp);
3853 3854 3855 3856 3857 3858

	if (cstate->minorversion) {
		unhash_stid(&stp->st_stid);
		free_generic_stateid(stp);
	} else
		oo->oo_last_closed_stid = stp;
3859

3860 3861 3862 3863 3864 3865 3866 3867 3868 3869
	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))
3870
				move_to_close_lru(oo, SVC_NET(rqstp));
3871 3872
		}
	}
L
Linus Torvalds 已提交
3873
out:
3874 3875
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3876 3877 3878
	return status;
}

3879
__be32
3880 3881
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3882
{
3883 3884
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3885
	struct nfs4_stid *s;
3886
	__be32 status;
3887
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3888

3889
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3890
		return status;
L
Linus Torvalds 已提交
3891 3892

	nfs4_lock_state();
3893 3894
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3895
	if (status)
3896
		goto out;
3897
	dp = delegstateid(s);
3898
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3899 3900 3901 3902
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3903
out:
3904 3905
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3906 3907 3908 3909 3910 3911
	return status;
}


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

3912
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3913

B
Benny Halevy 已提交
3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928
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;

3929
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3930 3931 3932 3933
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3934
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3935 3936 3937
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3938
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957
}

/*
 * 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;
}

3958 3959
/* 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. */
3960
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3961
};
L
Linus Torvalds 已提交
3962 3963 3964 3965

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3966
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3967

3968
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3969 3970 3971
		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);
3972 3973 3974
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3975 3976
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3977
	} else {
3978 3979 3980
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3981 3982
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3983 3984
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3985 3986
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3987 3988 3989 3990 3991 3992
		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
3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003
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;
}

4004 4005
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
4006
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4007
{
4008
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
4009
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4010

4011
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
4012 4013
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
4014 4015 4016 4017
	}
	return NULL;
}

4018
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
4019
{
4020 4021 4022
	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);
4023
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
4024

4025
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
4026
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
4027
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
4028 4029
}

L
Linus Torvalds 已提交
4030 4031 4032
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4033
 * occurred. 
L
Linus Torvalds 已提交
4034
 *
4035
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4036 4037
 */

4038
static struct nfs4_lockowner *
4039
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4040
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4041

4042 4043
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4044
		return NULL;
4045 4046
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4047 4048
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4049 4050 4051
	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
4052 4053
}

4054 4055
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
4056
{
4057
	struct nfs4_ol_stateid *stp;
4058
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4059

4060
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4061
	if (stp == NULL)
J
J. Bruce Fields 已提交
4062
		return NULL;
J
J. Bruce Fields 已提交
4063
	stp->st_stid.sc_type = NFS4_LOCK_STID;
4064
	list_add(&stp->st_perfile, &fp->fi_stateids);
4065 4066
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4067
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4068
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4069
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4070
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4071
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4072 4073 4074
	return stp;
}

4075
static int
L
Linus Torvalds 已提交
4076 4077
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4078
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4079 4080 4081
	     LOFF_OVERFLOW(offset, length)));
}

4082
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4083 4084 4085 4086
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4087
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4088 4089
		return;
	nfs4_file_get_access(fp, oflag);
4090
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4091 4092
}

J
J. Bruce Fields 已提交
4093
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)
4094 4095 4096 4097 4098 4099
{
	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;
4100
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4101

4102 4103
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4104 4105 4106 4107 4108 4109 4110 4111
	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;
	}
4112
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125
			&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
4126 4127 4128
/*
 *  LOCK operation 
 */
4129
__be32
4130
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4131
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4132
{
4133 4134
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4135
	struct nfs4_ol_stateid *lock_stp;
4136
	struct file *filp = NULL;
4137 4138
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4139
	__be32 status = 0;
4140
	bool new_state = false;
4141
	int lkflg;
4142
	int err;
4143 4144
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4145 4146 4147 4148 4149 4150 4151 4152

	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;

4153
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4154
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4155 4156 4157 4158
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4159 4160 4161
	nfs4_lock_state();

	if (lock->lk_is_new) {
4162
		struct nfs4_ol_stateid *open_stp = NULL;
4163 4164 4165 4166 4167 4168 4169

		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
4170
		status = nfserr_stale_clientid;
4171
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4172 4173 4174
			goto out;

		/* validate and update open stateid and open seqid */
4175
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4176 4177
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4178
					&open_stp, nn);
4179
		if (status)
L
Linus Torvalds 已提交
4180
			goto out;
4181
		open_sop = openowner(open_stp->st_stateowner);
4182
		status = nfserr_bad_stateid;
4183
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4184 4185
						&lock->v.new.clientid))
			goto out;
4186 4187
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4188
	} else
4189
		status = nfs4_preprocess_seqid_op(cstate,
4190 4191
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4192
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4193 4194
	if (status)
		goto out;
4195
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4196

4197 4198 4199 4200 4201
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4202
	status = nfserr_grace;
4203
	if (locks_in_grace(net) && !lock->lk_reclaim)
4204 4205
		goto out;
	status = nfserr_no_grace;
4206
	if (!locks_in_grace(net) && lock->lk_reclaim)
4207 4208
		goto out;

4209 4210 4211 4212 4213 4214 4215 4216
	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
4217 4218 4219
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4220 4221 4222
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4223
			file_lock->fl_type = F_RDLCK;
4224
			break;
L
Linus Torvalds 已提交
4225 4226
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4227 4228 4229
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4230
			file_lock->fl_type = F_WRLCK;
4231
			break;
L
Linus Torvalds 已提交
4232 4233 4234 4235
		default:
			status = nfserr_inval;
		goto out;
	}
4236 4237 4238 4239
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254
	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
4255

4256
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4257
	switch (-err) {
L
Linus Torvalds 已提交
4258
	case 0: /* success! */
4259 4260
		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
4261
				sizeof(stateid_t));
4262
		status = 0;
4263 4264 4265 4266
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4267
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4268
		break;
L
Linus Torvalds 已提交
4269 4270
	case (EDEADLK):
		status = nfserr_deadlock;
4271
		break;
4272
	default:
4273
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4274
		status = nfserrno(err);
4275
		break;
L
Linus Torvalds 已提交
4276 4277
	}
out:
4278
	if (status && new_state)
4279
		release_lockowner(lock_sop);
4280
	nfsd4_bump_seqid(cstate, status);
4281 4282
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4283 4284 4285 4286
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4287 4288 4289
	return status;
}

4290 4291 4292 4293 4294 4295
/*
 * 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.)
 */
4296
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4297 4298
{
	struct file *file;
4299 4300 4301 4302 4303
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4304 4305 4306
	return err;
}

L
Linus Torvalds 已提交
4307 4308 4309
/*
 * LOCKT operation
 */
4310
__be32
4311 4312
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4313 4314
{
	struct inode *inode;
4315
	struct file_lock *file_lock = NULL;
4316
	struct nfs4_lockowner *lo;
4317
	__be32 status;
4318
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4319

4320
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4321 4322 4323 4324 4325 4326 4327
		return nfserr_grace;

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

	nfs4_lock_state();

4328 4329 4330 4331 4332
	if (!nfsd4_has_session(cstate)) {
		status = lookup_clientid(&lockt->lt_clientid, false, nn, NULL);
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
4333

4334
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4335 4336
		goto out;

4337
	inode = cstate->current_fh.fh_dentry->d_inode;
4338 4339 4340 4341 4342 4343 4344
	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
4345 4346 4347
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4348
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4349 4350 4351
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4352
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4353 4354
		break;
		default:
4355
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4356 4357 4358 4359
			status = nfserr_inval;
		goto out;
	}

4360
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4361
	if (lo)
4362 4363 4364
		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
4365

4366 4367
	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
4368

4369
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4370

4371
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4372
	if (status)
4373
		goto out;
4374

4375
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4376
		status = nfserr_denied;
4377
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4378 4379 4380
	}
out:
	nfs4_unlock_state();
4381 4382
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4383 4384 4385
	return status;
}

4386
__be32
4387
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4388
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4389
{
4390
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4391
	struct file *filp = NULL;
4392
	struct file_lock *file_lock = NULL;
4393
	__be32 status;
4394
	int err;
4395 4396
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4397 4398 4399 4400 4401 4402 4403 4404 4405
	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();
									        
4406
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4407 4408
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4409
	if (status)
L
Linus Torvalds 已提交
4410
		goto out;
4411 4412 4413 4414 4415
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
	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
4434

4435
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4436
	if (err) {
4437
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4438 4439
		goto out_nfserr;
	}
4440 4441
	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
4442 4443

out:
4444
	nfsd4_bump_seqid(cstate, status);
4445 4446
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4447 4448
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4449 4450 4451
	return status;

out_nfserr:
4452
	status = nfserrno(err);
L
Linus Torvalds 已提交
4453 4454 4455 4456 4457 4458 4459 4460 4461
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4462
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4463 4464
{
	struct file_lock **flpp;
4465
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4466 4467
	int status = 0;

4468
	lock_flocks();
L
Linus Torvalds 已提交
4469
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4470
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4471 4472
			status = 1;
			goto out;
4473
		}
L
Linus Torvalds 已提交
4474 4475
	}
out:
4476
	unlock_flocks();
L
Linus Torvalds 已提交
4477 4478 4479
	return status;
}

4480
__be32
4481 4482 4483
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4484 4485
{
	clientid_t *clid = &rlockowner->rl_clientid;
4486
	struct nfs4_stateowner *sop;
4487
	struct nfs4_lockowner *lo;
4488
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4489
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4490
	struct list_head matches;
4491
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4492
	__be32 status;
4493
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4494 4495 4496 4497 4498 4499

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

	nfs4_lock_state();

4500 4501 4502 4503
	status = lookup_clientid(clid, cstate->minorversion, nn, NULL);
	if (status)
		goto out;

4504 4505
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4506

4507
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4508 4509
		if (sop->so_is_open_owner)
			continue;
4510 4511 4512 4513 4514 4515 4516 4517
		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
4518
		}
4519 4520 4521 4522 4523
	}
	/* 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
4524
	while (!list_empty(&matches)) {
4525 4526
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4527 4528
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4529 4530
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4531 4532 4533 4534 4535 4536 4537
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4538
alloc_reclaim(void)
L
Linus Torvalds 已提交
4539
{
N
NeilBrown 已提交
4540
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4541 4542
}

4543
bool
4544
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4545
{
4546
	struct nfs4_client_reclaim *crp;
4547

4548
	crp = nfsd4_find_reclaim_client(name, nn);
4549
	return (crp && crp->cr_clp);
4550 4551
}

L
Linus Torvalds 已提交
4552 4553 4554
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4555
struct nfs4_client_reclaim *
4556
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4557 4558
{
	unsigned int strhashval;
4559
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4560

N
NeilBrown 已提交
4561 4562
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4563 4564 4565
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4566
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4567
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4568
		crp->cr_clp = NULL;
4569
		nn->reclaim_str_hashtbl_size++;
4570 4571
	}
	return crp;
L
Linus Torvalds 已提交
4572 4573
}

4574
void
4575
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4576 4577 4578
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4579
	nn->reclaim_str_hashtbl_size--;
4580 4581
}

4582
void
4583
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4584 4585 4586 4587 4588
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4589 4590
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4591
			                struct nfs4_client_reclaim, cr_strhash);
4592
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4593 4594
		}
	}
4595
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4596 4597 4598 4599
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4600
struct nfs4_client_reclaim *
4601
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4602 4603 4604 4605
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4606
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4607

4608
	strhashval = clientstr_hashval(recdir);
4609
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4610
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4611 4612 4613 4614 4615 4616 4617 4618 4619
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4620
__be32
4621
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4622
{
4623 4624 4625
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4626
	clp = find_confirmed_client(clid, sessions, nn);
4627 4628 4629 4630
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4631 4632
}

B
Bryan Schumaker 已提交
4633 4634
#ifdef CONFIG_NFSD_FAULT_INJECTION

4635 4636
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
4637 4638
	if (mark_client_expired(clp))
		return 0;
4639 4640 4641 4642
	expire_client(clp);
	return 1;
}

4643 4644 4645
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
4646
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4647 4648 4649 4650 4651 4652 4653 4654
	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];
4655
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4656 4657 4658
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

B
Bryan Schumaker 已提交
4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684
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);
}

4685 4686 4687 4688 4689 4690 4691
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;
}

4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711
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);
}

4712 4713 4714 4715 4716 4717 4718
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;
}

4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764
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;
}

4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776
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;
}

4777
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4778 4779
{
	struct nfs4_client *clp, *next;
4780
	u64 count = 0;
4781
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4782

4783 4784 4785
	if (!nfsd_netns_ready(nn))
		return 0;

4786
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4787 4788
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4789 4790 4791
			break;
	}

4792 4793 4794
	return count;
}

4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809
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
4810 4811
#endif /* CONFIG_NFSD_FAULT_INJECTION */

4812
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4813

4814
void
4815
nfs4_state_init(void)
L
Linus Torvalds 已提交
4816
{
4817 4818
}

4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839
/*
 * 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);
}

4840
static int nfs4_state_create_net(struct net *net)
4841 4842 4843 4844 4845 4846 4847
{
	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)
4848
		goto err;
4849 4850 4851 4852
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4853 4854 4855 4856
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4857 4858 4859 4860
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4861 4862 4863 4864
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4865

4866
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4867
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4868
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4869
	}
4870 4871
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4872 4873
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4874 4875
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4876
	nn->conf_name_tree = RB_ROOT;
4877
	nn->unconf_name_tree = RB_ROOT;
4878
	INIT_LIST_HEAD(&nn->client_lru);
4879
	INIT_LIST_HEAD(&nn->close_lru);
4880
	INIT_LIST_HEAD(&nn->del_recall_lru);
4881
	spin_lock_init(&nn->client_lock);
4882

4883
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4884
	get_net(net);
4885

4886
	return 0;
4887

4888 4889
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4890 4891
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4892 4893
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4894 4895
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4896 4897
err:
	return -ENOMEM;
4898 4899 4900
}

static void
4901
nfs4_state_destroy_net(struct net *net)
4902 4903 4904 4905
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4906
	struct rb_node *node, *tmp;
4907 4908 4909 4910 4911 4912 4913

	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);
		}
	}
4914 4915 4916 4917 4918 4919 4920 4921 4922 4923

	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);
	}

4924
	kfree(nn->sessionid_hashtbl);
4925
	kfree(nn->lockowner_ino_hashtbl);
4926
	kfree(nn->ownerstr_hashtbl);
4927
	kfree(nn->unconf_id_hashtbl);
4928
	kfree(nn->conf_id_hashtbl);
4929
	put_net(net);
4930 4931
}

4932
int
4933
nfs4_state_start_net(struct net *net)
4934
{
4935
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4936 4937
	int ret;

4938
	ret = nfs4_state_create_net(net);
4939 4940
	if (ret)
		return ret;
4941
	nfsd4_client_tracking_init(net);
4942
	nn->boot_time = get_seconds();
4943
	locks_start_grace(net, &nn->nfsd4_manager);
4944
	nn->grace_ended = false;
4945
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4946 4947
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4948 4949 4950 4951 4952 4953 4954 4955 4956 4957
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

4958
	ret = set_callback_cred();
4959 4960
	if (ret)
		return -ENOMEM;
4961
	laundry_wq = create_singlethread_workqueue("nfsd4");
4962 4963 4964 4965
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4966 4967 4968
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4969

4970
	set_max_delegations();
4971

4972
	return 0;
4973

4974 4975
out_free_laundry:
	destroy_workqueue(laundry_wq);
4976
out_recovery:
4977
	return ret;
L
Linus Torvalds 已提交
4978 4979
}

4980
/* should be called with the state lock held */
4981
void
4982
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4983 4984 4985
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4986
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4987

4988 4989
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4990

L
Linus Torvalds 已提交
4991 4992
	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
4993
	list_for_each_safe(pos, next, &nn->del_recall_lru) {
L
Linus Torvalds 已提交
4994 4995 4996 4997 4998 4999 5000 5001 5002
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
		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);
	}

5003
	nfsd4_client_tracking_exit(net);
5004
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
5005 5006 5007 5008 5009
}

void
nfs4_state_shutdown(void)
{
5010
	destroy_workqueue(laundry_wq);
5011
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
5012
}
5013 5014 5015 5016

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5017 5018
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
5019 5020 5021 5022 5023
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5024 5025 5026 5027 5028 5029 5030 5031 5032 5033
	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);
5034 5035
}

5036 5037 5038
/*
 * functions to set current state id
 */
5039 5040 5041 5042 5043 5044
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5045 5046 5047 5048 5049 5050
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065
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
 */
5066

5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078
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);
}

5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090
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);
}

5091 5092 5093 5094 5095 5096 5097
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5098
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5099
{
5100
	get_stateid(cstate, &locku->lu_stateid);
5101
}
5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113

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);
}