nfs4state.c 150.3 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 <linux/hash.h>
45
#include "xdr4.h"
46
#include "xdr4cb.h"
47
#include "vfs.h"
48
#include "current_stateid.h"
L
Linus Torvalds 已提交
49

50 51
#include "netns.h"

L
Linus Torvalds 已提交
52 53
#define NFSDDBG_FACILITY                NFSDDBG_PROC

54 55 56 57 58 59 60 61
#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 */
};
62 63 64
static const stateid_t currentstateid = {
	.si_generation = 1,
};
65

A
Andy Adamson 已提交
66
static u64 current_sessionid = 1;
67

68 69
#define ZERO_STATEID(stateid) (!memcmp((stateid), &zero_stateid, sizeof(stateid_t)))
#define ONE_STATEID(stateid)  (!memcmp((stateid), &one_stateid, sizeof(stateid_t)))
70
#define CURRENT_STATEID(stateid) (!memcmp((stateid), &currentstateid, sizeof(stateid_t)))
L
Linus Torvalds 已提交
71 72

/* forward declarations */
73
static bool check_for_locks(struct nfs4_file *fp, struct nfs4_lockowner *lowner);
74
static void nfs4_free_ol_stateid(struct nfs4_stid *stid);
L
Linus Torvalds 已提交
75

76 77 78
/* Locking: */

/* Currently used for almost all code touching nfsv4 state: */
I
Ingo Molnar 已提交
79
static DEFINE_MUTEX(client_mutex);
L
Linus Torvalds 已提交
80

81 82 83 84 85
/*
 * 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:
 */
86
static DEFINE_SPINLOCK(state_lock);
87

C
Christoph Hellwig 已提交
88 89 90 91 92
static struct kmem_cache *openowner_slab;
static struct kmem_cache *lockowner_slab;
static struct kmem_cache *file_slab;
static struct kmem_cache *stateid_slab;
static struct kmem_cache *deleg_slab;
N
NeilBrown 已提交
93

L
Linus Torvalds 已提交
94 95 96
void
nfs4_lock_state(void)
{
I
Ingo Molnar 已提交
97
	mutex_lock(&client_mutex);
L
Linus Torvalds 已提交
98 99
}

100
static void free_session(struct nfsd4_session *);
101

102
static bool is_session_dead(struct nfsd4_session *ses)
103
{
104
	return ses->se_flags & NFS4_SESSION_DEAD;
105 106
}

107
static __be32 mark_session_dead_locked(struct nfsd4_session *ses, int ref_held_by_me)
108
{
109
	if (atomic_read(&ses->se_ref) > ref_held_by_me)
110 111 112
		return nfserr_jukebox;
	ses->se_flags |= NFS4_SESSION_DEAD;
	return nfs_ok;
113 114
}

L
Linus Torvalds 已提交
115 116 117
void
nfs4_unlock_state(void)
{
I
Ingo Molnar 已提交
118
	mutex_unlock(&client_mutex);
L
Linus Torvalds 已提交
119 120
}

121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
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);
}

185
static void put_client_renew_locked(struct nfs4_client *clp)
186 187 188 189 190 191 192
{
	if (!atomic_dec_and_test(&clp->cl_refcount))
		return;
	if (!is_client_expired(clp))
		renew_client_locked(clp);
}

193 194 195 196
static void put_client_renew(struct nfs4_client *clp)
{
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

197 198 199 200
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
		return;
	if (!is_client_expired(clp))
		renew_client_locked(clp);
201 202 203
	spin_unlock(&nn->client_lock);
}

204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
static __be32 nfsd4_get_session_locked(struct nfsd4_session *ses)
{
	__be32 status;

	if (is_session_dead(ses))
		return nfserr_badsession;
	status = get_client_locked(ses->se_client);
	if (status)
		return status;
	atomic_inc(&ses->se_ref);
	return nfs_ok;
}

static void nfsd4_put_session_locked(struct nfsd4_session *ses)
{
	struct nfs4_client *clp = ses->se_client;

	if (atomic_dec_and_test(&ses->se_ref) && is_session_dead(ses))
		free_session(ses);
	put_client_renew_locked(clp);
}

static void nfsd4_put_session(struct nfsd4_session *ses)
{
	struct nfs4_client *clp = ses->se_client;
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
	nfsd4_put_session_locked(ses);
	spin_unlock(&nn->client_lock);
}


L
Linus Torvalds 已提交
237 238 239 240 241 242 243 244 245 246 247 248 249
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;
}

250 251 252 253 254
static void nfsd4_free_file(struct nfs4_file *f)
{
	kmem_cache_free(file_slab, f);
}

255 256 257
static inline void
put_nfs4_file(struct nfs4_file *fi)
{
258 259
	might_lock(&state_lock);

260
	if (atomic_dec_and_lock(&fi->fi_ref, &state_lock)) {
261
		hlist_del(&fi->fi_hash);
262
		spin_unlock(&state_lock);
263
		nfsd4_free_file(fi);
264
	}
265 266 267 268 269
}

static inline void
get_nfs4_file(struct nfs4_file *fi)
{
270
	atomic_inc(&fi->fi_ref);
271 272
}

273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345
static struct file *
__nfs4_get_fd(struct nfs4_file *f, int oflag)
{
	if (f->fi_fds[oflag])
		return get_file(f->fi_fds[oflag]);
	return NULL;
}

static struct file *
find_writeable_file_locked(struct nfs4_file *f)
{
	struct file *ret;

	lockdep_assert_held(&f->fi_lock);

	ret = __nfs4_get_fd(f, O_WRONLY);
	if (!ret)
		ret = __nfs4_get_fd(f, O_RDWR);
	return ret;
}

static struct file *
find_writeable_file(struct nfs4_file *f)
{
	struct file *ret;

	spin_lock(&f->fi_lock);
	ret = find_writeable_file_locked(f);
	spin_unlock(&f->fi_lock);

	return ret;
}

static struct file *find_readable_file_locked(struct nfs4_file *f)
{
	struct file *ret;

	lockdep_assert_held(&f->fi_lock);

	ret = __nfs4_get_fd(f, O_RDONLY);
	if (!ret)
		ret = __nfs4_get_fd(f, O_RDWR);
	return ret;
}

static struct file *
find_readable_file(struct nfs4_file *f)
{
	struct file *ret;

	spin_lock(&f->fi_lock);
	ret = find_readable_file_locked(f);
	spin_unlock(&f->fi_lock);

	return ret;
}

static struct file *
find_any_file(struct nfs4_file *f)
{
	struct file *ret;

	spin_lock(&f->fi_lock);
	ret = __nfs4_get_fd(f, O_RDWR);
	if (!ret) {
		ret = __nfs4_get_fd(f, O_WRONLY);
		if (!ret)
			ret = __nfs4_get_fd(f, O_RDONLY);
	}
	spin_unlock(&f->fi_lock);
	return ret;
}

346
static atomic_long_t num_delegations;
347
unsigned long max_delegations;
348 349 350 351 352

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

353 354 355 356
/* 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)
357

358
static unsigned int ownerstr_hashval(u32 clientid, struct xdr_netobj *ownername)
359 360 361 362 363
{
	unsigned int ret;

	ret = opaque_hashval(ownername->data, ownername->len);
	ret += clientid;
364
	return ret & OWNER_HASH_MASK;
365
}
366 367 368 369

/* hash table for nfs4_file */
#define FILE_HASH_BITS                   8
#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
S
Shan Wei 已提交
370

371
static unsigned int nfsd_fh_hashval(struct knfsd_fh *fh)
372
{
373 374 375 376 377 378 379 380 381 382 383 384 385 386
	return jhash2(fh->fh_base.fh_pad, XDR_QUADLEN(fh->fh_size), 0);
}

static unsigned int file_hashval(struct knfsd_fh *fh)
{
	return nfsd_fh_hashval(fh) & (FILE_HASH_SIZE - 1);
}

static bool nfsd_fh_match(struct knfsd_fh *fh1, struct knfsd_fh *fh2)
{
	return fh1->fh_size == fh2->fh_size &&
		!memcmp(fh1->fh_base.fh_pad,
				fh2->fh_base.fh_pad,
				fh1->fh_size);
387 388
}

389
static struct hlist_head file_hashtbl[FILE_HASH_SIZE];
390

391 392
static void
__nfs4_file_get_access(struct nfs4_file *fp, u32 access)
393
{
394 395
	lockdep_assert_held(&fp->fi_lock);

396 397 398 399
	if (access & NFS4_SHARE_ACCESS_WRITE)
		atomic_inc(&fp->fi_access[O_WRONLY]);
	if (access & NFS4_SHARE_ACCESS_READ)
		atomic_inc(&fp->fi_access[O_RDONLY]);
400 401
}

402 403
static __be32
nfs4_file_get_access(struct nfs4_file *fp, u32 access)
404
{
405 406
	lockdep_assert_held(&fp->fi_lock);

407 408 409 410
	/* Does this access mode make sense? */
	if (access & ~NFS4_SHARE_ACCESS_BOTH)
		return nfserr_inval;

411 412 413 414
	/* Does it conflict with a deny mode already set? */
	if ((access & fp->fi_share_deny) != 0)
		return nfserr_share_denied;

415 416
	__nfs4_file_get_access(fp, access);
	return nfs_ok;
417 418
}

419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
static __be32 nfs4_file_check_deny(struct nfs4_file *fp, u32 deny)
{
	/* Common case is that there is no deny mode. */
	if (deny) {
		/* Does this deny mode make sense? */
		if (deny & ~NFS4_SHARE_DENY_BOTH)
			return nfserr_inval;

		if ((deny & NFS4_SHARE_DENY_READ) &&
		    atomic_read(&fp->fi_access[O_RDONLY]))
			return nfserr_share_denied;

		if ((deny & NFS4_SHARE_DENY_WRITE) &&
		    atomic_read(&fp->fi_access[O_WRONLY]))
			return nfserr_share_denied;
	}
	return nfs_ok;
}

438
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
439
{
440 441 442 443 444 445
	might_lock(&fp->fi_lock);

	if (atomic_dec_and_lock(&fp->fi_access[oflag], &fp->fi_lock)) {
		struct file *f1 = NULL;
		struct file *f2 = NULL;

J
Jeff Layton 已提交
446
		swap(f1, fp->fi_fds[oflag]);
447
		if (atomic_read(&fp->fi_access[1 - oflag]) == 0)
J
Jeff Layton 已提交
448
			swap(f2, fp->fi_fds[O_RDWR]);
449 450 451 452 453
		spin_unlock(&fp->fi_lock);
		if (f1)
			fput(f1);
		if (f2)
			fput(f2);
454 455 456
	}
}

457
static void nfs4_file_put_access(struct nfs4_file *fp, u32 access)
458
{
459 460 461
	WARN_ON_ONCE(access & ~NFS4_SHARE_ACCESS_BOTH);

	if (access & NFS4_SHARE_ACCESS_WRITE)
462
		__nfs4_file_put_access(fp, O_WRONLY);
463 464
	if (access & NFS4_SHARE_ACCESS_READ)
		__nfs4_file_put_access(fp, O_RDONLY);
465 466
}

467 468
static struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl,
					 struct kmem_cache *slab)
469
{
J
J. Bruce Fields 已提交
470
	struct nfs4_stid *stid;
J
J. Bruce Fields 已提交
471
	int new_id;
472

473
	stid = kmem_cache_zalloc(slab, GFP_KERNEL);
J
J. Bruce Fields 已提交
474 475 476
	if (!stid)
		return NULL;

477 478 479 480 481
	idr_preload(GFP_KERNEL);
	spin_lock(&cl->cl_lock);
	new_id = idr_alloc_cyclic(&cl->cl_stateids, stid, 0, 0, GFP_NOWAIT);
	spin_unlock(&cl->cl_lock);
	idr_preload_end();
T
Tejun Heo 已提交
482
	if (new_id < 0)
J
J. Bruce Fields 已提交
483
		goto out_free;
484
	stid->sc_client = cl;
J
J. Bruce Fields 已提交
485 486
	stid->sc_stateid.si_opaque.so_id = new_id;
	stid->sc_stateid.si_opaque.so_clid = cl->cl_clientid;
487
	/* Will be incremented before return to client: */
488
	atomic_set(&stid->sc_count, 1);
489 490

	/*
J
J. Bruce Fields 已提交
491 492 493 494 495 496 497
	 * 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):
498
	 */
J
J. Bruce Fields 已提交
499 500
	return stid;
out_free:
501
	kmem_cache_free(slab, stid);
J
J. Bruce Fields 已提交
502
	return NULL;
503 504
}

505
static struct nfs4_ol_stateid * nfs4_alloc_open_stateid(struct nfs4_client *clp)
506
{
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
	struct nfs4_stid *stid;
	struct nfs4_ol_stateid *stp;

	stid = nfs4_alloc_stid(clp, stateid_slab);
	if (!stid)
		return NULL;

	stp = openlockstateid(stid);
	stp->st_stid.sc_free = nfs4_free_ol_stateid;
	return stp;
}

static void nfs4_free_deleg(struct nfs4_stid *stid)
{
	kmem_cache_free(deleg_slab, stid);
	atomic_long_dec(&num_delegations);
523 524
}

525 526 527 528 529 530 531 532 533 534 535 536 537 538
/*
 * When we recall a delegation, we should be careful not to hand it
 * out again straight away.
 * To ensure this we keep a pair of bloom filters ('new' and 'old')
 * in which the filehandles of recalled delegations are "stored".
 * If a filehandle appear in either filter, a delegation is blocked.
 * When a delegation is recalled, the filehandle is stored in the "new"
 * filter.
 * Every 30 seconds we swap the filters and clear the "new" one,
 * unless both are empty of course.
 *
 * Each filter is 256 bits.  We hash the filehandle to 32bit and use the
 * low 3 bytes as hash-table indices.
 *
539
 * 'blocked_delegations_lock', which is always taken in block_delegations(),
540 541 542
 * is used to manage concurrent access.  Testing does not need the lock
 * except when swapping the two filters.
 */
543
static DEFINE_SPINLOCK(blocked_delegations_lock);
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558
static struct bloom_pair {
	int	entries, old_entries;
	time_t	swap_time;
	int	new; /* index into 'set' */
	DECLARE_BITMAP(set[2], 256);
} blocked_delegations;

static int delegation_blocked(struct knfsd_fh *fh)
{
	u32 hash;
	struct bloom_pair *bd = &blocked_delegations;

	if (bd->entries == 0)
		return 0;
	if (seconds_since_boot() - bd->swap_time > 30) {
559
		spin_lock(&blocked_delegations_lock);
560 561 562 563 564 565 566 567
		if (seconds_since_boot() - bd->swap_time > 30) {
			bd->entries -= bd->old_entries;
			bd->old_entries = bd->entries;
			memset(bd->set[bd->new], 0,
			       sizeof(bd->set[0]));
			bd->new = 1-bd->new;
			bd->swap_time = seconds_since_boot();
		}
568
		spin_unlock(&blocked_delegations_lock);
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590
	}
	hash = arch_fast_hash(&fh->fh_base, fh->fh_size, 0);
	if (test_bit(hash&255, bd->set[0]) &&
	    test_bit((hash>>8)&255, bd->set[0]) &&
	    test_bit((hash>>16)&255, bd->set[0]))
		return 1;

	if (test_bit(hash&255, bd->set[1]) &&
	    test_bit((hash>>8)&255, bd->set[1]) &&
	    test_bit((hash>>16)&255, bd->set[1]))
		return 1;

	return 0;
}

static void block_delegations(struct knfsd_fh *fh)
{
	u32 hash;
	struct bloom_pair *bd = &blocked_delegations;

	hash = arch_fast_hash(&fh->fh_base, fh->fh_size, 0);

591
	spin_lock(&blocked_delegations_lock);
592 593 594 595 596 597
	__set_bit(hash&255, bd->set[bd->new]);
	__set_bit((hash>>8)&255, bd->set[bd->new]);
	__set_bit((hash>>16)&255, bd->set[bd->new]);
	if (bd->entries == 0)
		bd->swap_time = seconds_since_boot();
	bd->entries += 1;
598
	spin_unlock(&blocked_delegations_lock);
599 600
}

L
Linus Torvalds 已提交
601
static struct nfs4_delegation *
602
alloc_init_deleg(struct nfs4_client *clp, struct svc_fh *current_fh)
L
Linus Torvalds 已提交
603 604
{
	struct nfs4_delegation *dp;
605
	long n;
L
Linus Torvalds 已提交
606 607

	dprintk("NFSD alloc_init_deleg\n");
608 609 610
	n = atomic_long_inc_return(&num_delegations);
	if (n < 0 || n > max_delegations)
		goto out_dec;
611
	if (delegation_blocked(&current_fh->fh_handle))
612
		goto out_dec;
613
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
614
	if (dp == NULL)
615
		goto out_dec;
616 617

	dp->dl_stid.sc_free = nfs4_free_deleg;
618 619
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
620 621
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
622 623
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
624 625
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
626
	INIT_LIST_HEAD(&dp->dl_recall_lru);
627
	dp->dl_type = NFS4_OPEN_DELEGATE_READ;
628
	INIT_WORK(&dp->dl_recall.cb_work, nfsd4_run_cb_recall);
L
Linus Torvalds 已提交
629
	return dp;
630 631 632
out_dec:
	atomic_long_dec(&num_delegations);
	return NULL;
L
Linus Torvalds 已提交
633 634 635
}

void
636
nfs4_put_stid(struct nfs4_stid *s)
L
Linus Torvalds 已提交
637
{
638
	struct nfs4_file *fp = s->sc_file;
639 640
	struct nfs4_client *clp = s->sc_client;

641 642 643
	might_lock(&clp->cl_lock);

	if (!atomic_dec_and_lock(&s->sc_count, &clp->cl_lock))
644 645
		return;
	idr_remove(&clp->cl_stateids, s->sc_stateid.si_opaque.so_id);
646
	spin_unlock(&clp->cl_lock);
647
	s->sc_free(s);
648 649
	if (fp)
		put_nfs4_file(fp);
L
Linus Torvalds 已提交
650 651
}

652
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
653
{
654 655
	lockdep_assert_held(&state_lock);

656 657
	if (!fp->fi_lease)
		return;
658 659 660
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
661
		fput(fp->fi_deleg_file);
662 663
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
664 665
}

J
J. Bruce Fields 已提交
666 667
static void unhash_stid(struct nfs4_stid *s)
{
J
J. Bruce Fields 已提交
668
	s->sc_type = 0;
J
J. Bruce Fields 已提交
669 670
}

671 672 673
static void
hash_delegation_locked(struct nfs4_delegation *dp, struct nfs4_file *fp)
{
674
	lockdep_assert_held(&state_lock);
675
	lockdep_assert_held(&fp->fi_lock);
676

677
	dp->dl_stid.sc_type = NFS4_DELEG_STID;
678 679 680 681
	list_add(&dp->dl_perfile, &fp->fi_delegations);
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
}

L
Linus Torvalds 已提交
682
static void
683
unhash_delegation_locked(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
684
{
685
	struct nfs4_file *fp = dp->dl_stid.sc_file;
686

687 688
	lockdep_assert_held(&state_lock);

689
	dp->dl_stid.sc_type = NFS4_CLOSED_DELEG_STID;
690 691
	/* Ensure that deleg break won't try to requeue it */
	++dp->dl_time;
692
	spin_lock(&fp->fi_lock);
693
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
694
	list_del_init(&dp->dl_recall_lru);
695 696
	list_del_init(&dp->dl_perfile);
	spin_unlock(&fp->fi_lock);
697
	if (fp)
698
		nfs4_put_deleg_lease(fp);
699 700 701 702
}

static void destroy_delegation(struct nfs4_delegation *dp)
{
703 704 705
	spin_lock(&state_lock);
	unhash_delegation_locked(dp);
	spin_unlock(&state_lock);
706
	nfs4_put_stid(&dp->dl_stid);
707 708 709 710 711 712
}

static void revoke_delegation(struct nfs4_delegation *dp)
{
	struct nfs4_client *clp = dp->dl_stid.sc_client;

713 714
	WARN_ON(!list_empty(&dp->dl_recall_lru));

715
	if (clp->cl_minorversion == 0)
716
		nfs4_put_stid(&dp->dl_stid);
717 718
	else {
		dp->dl_stid.sc_type = NFS4_REVOKED_DELEG_STID;
719 720 721
		spin_lock(&clp->cl_lock);
		list_add(&dp->dl_recall_lru, &clp->cl_revoked);
		spin_unlock(&clp->cl_lock);
722 723 724
	}
}

L
Linus Torvalds 已提交
725 726 727 728
/* 
 * SETCLIENTID state 
 */

729 730 731 732 733 734 735 736 737 738
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;
}

739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756
/*
 * 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.
 */
757 758
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
759
	int i;
760
	unsigned int access = 0;
761 762 763

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
764
			access |= i;
765
	}
766
	return access;
767 768
}

769 770 771 772
/* set share access for a given stateid */
static inline void
set_access(u32 access, struct nfs4_ol_stateid *stp)
{
773 774 775 776
	unsigned char mask = 1 << access;

	WARN_ON_ONCE(access > NFS4_SHARE_ACCESS_BOTH);
	stp->st_access_bmap |= mask;
777 778 779 780 781 782
}

/* clear share access for a given stateid */
static inline void
clear_access(u32 access, struct nfs4_ol_stateid *stp)
{
783 784 785 786
	unsigned char mask = 1 << access;

	WARN_ON_ONCE(access > NFS4_SHARE_ACCESS_BOTH);
	stp->st_access_bmap &= ~mask;
787 788 789 790 791 792
}

/* test whether a given stateid has access */
static inline bool
test_access(u32 access, struct nfs4_ol_stateid *stp)
{
793 794 795
	unsigned char mask = 1 << access;

	return (bool)(stp->st_access_bmap & mask);
796 797
}

798 799
/* set share deny for a given stateid */
static inline void
800
set_deny(u32 deny, struct nfs4_ol_stateid *stp)
801
{
802 803 804 805
	unsigned char mask = 1 << deny;

	WARN_ON_ONCE(deny > NFS4_SHARE_DENY_BOTH);
	stp->st_deny_bmap |= mask;
806 807 808 809
}

/* clear share deny for a given stateid */
static inline void
810
clear_deny(u32 deny, struct nfs4_ol_stateid *stp)
811
{
812 813 814 815
	unsigned char mask = 1 << deny;

	WARN_ON_ONCE(deny > NFS4_SHARE_DENY_BOTH);
	stp->st_deny_bmap &= ~mask;
816 817 818 819
}

/* test whether a given stateid is denying specific access */
static inline bool
820
test_deny(u32 deny, struct nfs4_ol_stateid *stp)
821
{
822 823 824
	unsigned char mask = 1 << deny;

	return (bool)(stp->st_deny_bmap & mask);
825 826 827 828
}

static int nfs4_access_to_omode(u32 access)
{
829
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
830 831 832 833 834 835 836
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
837 838
	WARN_ON_ONCE(1);
	return O_RDONLY;
839 840
}

841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
/*
 * A stateid that had a deny mode associated with it is being released
 * or downgraded. Recalculate the deny mode on the file.
 */
static void
recalculate_deny_mode(struct nfs4_file *fp)
{
	struct nfs4_ol_stateid *stp;

	spin_lock(&fp->fi_lock);
	fp->fi_share_deny = 0;
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile)
		fp->fi_share_deny |= bmap_to_share_mode(stp->st_deny_bmap);
	spin_unlock(&fp->fi_lock);
}

static void
reset_union_bmap_deny(u32 deny, struct nfs4_ol_stateid *stp)
{
	int i;
	bool change = false;

	for (i = 1; i < 4; i++) {
		if ((i & deny) != i) {
			change = true;
			clear_deny(i, stp);
		}
	}

	/* Recalculate per-file deny mode if there was a change */
	if (change)
872
		recalculate_deny_mode(stp->st_stid.sc_file);
873 874
}

875 876 877 878 879
/* release all access and file references for a given stateid */
static void
release_all_access(struct nfs4_ol_stateid *stp)
{
	int i;
880
	struct nfs4_file *fp = stp->st_stid.sc_file;
881 882 883

	if (fp && stp->st_deny_bmap != 0)
		recalculate_deny_mode(fp);
884 885 886

	for (i = 1; i < 4; i++) {
		if (test_access(i, stp))
887
			nfs4_file_put_access(stp->st_stid.sc_file, i);
888 889 890 891
		clear_access(i, stp);
	}
}

892
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
893
{
894
	struct nfs4_file *fp = stp->st_stid.sc_file;
895

896 897
	lockdep_assert_held(&stp->st_stateowner->so_client->cl_lock);

898
	spin_lock(&fp->fi_lock);
899
	list_del(&stp->st_perfile);
900
	spin_unlock(&fp->fi_lock);
901 902 903
	list_del(&stp->st_perstateowner);
}

904
static void nfs4_free_ol_stateid(struct nfs4_stid *stid)
905
{
906
	struct nfs4_ol_stateid *stp = openlockstateid(stid);
907

908 909
	release_all_access(stp);
	kmem_cache_free(stateid_slab, stid);
910 911
}

912
static void nfs4_free_lock_stateid(struct nfs4_stid *stid)
913
{
914 915
	struct nfs4_ol_stateid *stp = openlockstateid(stid);
	struct nfs4_lockowner *lo = lockowner(stp->st_stateowner);
916 917
	struct file *file;

918 919 920 921 922 923 924 925
	file = find_any_file(stp->st_stid.sc_file);
	if (file)
		filp_close(file, (fl_owner_t)lo);
	nfs4_free_ol_stateid(stid);
}

static void __release_lock_stateid(struct nfs4_ol_stateid *stp)
{
926 927 928
	struct nfs4_openowner *oo = openowner(stp->st_openstp->st_stateowner);

	spin_lock(&oo->oo_owner.so_client->cl_lock);
929
	list_del(&stp->st_locks);
930
	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
931
	unhash_stid(&stp->st_stid);
932
	spin_unlock(&oo->oo_owner.so_client->cl_lock);
933
	nfs4_put_stid(&stp->st_stid);
934 935
}

936
static void unhash_lockowner(struct nfs4_lockowner *lo)
937
{
938
	struct nfs4_ol_stateid *stp;
939

940 941 942
	list_del(&lo->lo_owner.so_strhash);
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
943
				struct nfs4_ol_stateid, st_perstateowner);
944
		__release_lock_stateid(stp);
945 946 947
	}
}

948 949 950 951 952 953
static void nfs4_free_lockowner(struct nfs4_lockowner *lo)
{
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
}

954
static void release_lockowner(struct nfs4_lockowner *lo)
955
{
956 957
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
958 959
}

960 961 962 963 964 965 966
static void release_lockowner_if_empty(struct nfs4_lockowner *lo)
{
	if (list_empty(&lo->lo_owner.so_stateids))
		release_lockowner(lo);
}

static void release_lock_stateid(struct nfs4_ol_stateid *stp)
967
{
968
	struct nfs4_lockowner *lo;
969

970 971 972 973 974 975
	lo = lockowner(stp->st_stateowner);
	__release_lock_stateid(stp);
	release_lockowner_if_empty(lo);
}

static void release_open_stateid_locks(struct nfs4_ol_stateid *open_stp)
976 977
	__releases(&open_stp->st_stateowner->so_client->cl_lock)
	__acquires(&open_stp->st_stateowner->so_client->cl_lock)
978 979 980 981 982 983
{
	struct nfs4_ol_stateid *stp;

	while (!list_empty(&open_stp->st_locks)) {
		stp = list_entry(open_stp->st_locks.next,
				struct nfs4_ol_stateid, st_locks);
984
		spin_unlock(&open_stp->st_stateowner->so_client->cl_lock);
985
		release_lock_stateid(stp);
986
		spin_lock(&open_stp->st_stateowner->so_client->cl_lock);
987 988 989
	}
}

990
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
991
{
992
	spin_lock(&stp->st_stateowner->so_client->cl_lock);
993
	unhash_generic_stateid(stp);
994
	release_open_stateid_locks(stp);
995
	spin_unlock(&stp->st_stateowner->so_client->cl_lock);
996 997 998 999 1000
}

static void release_open_stateid(struct nfs4_ol_stateid *stp)
{
	unhash_open_stateid(stp);
1001
	nfs4_put_stid(&stp->st_stid);
1002 1003
}

1004
static void unhash_openowner(struct nfs4_openowner *oo)
1005
{
1006
	struct nfs4_ol_stateid *stp;
1007

1008 1009 1010 1011
	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,
1012
				struct nfs4_ol_stateid, st_perstateowner);
1013
		release_open_stateid(stp);
1014 1015 1016
	}
}

1017 1018 1019 1020 1021
static void release_last_closed_stateid(struct nfs4_openowner *oo)
{
	struct nfs4_ol_stateid *s = oo->oo_last_closed_stid;

	if (s) {
1022
		nfs4_put_stid(&s->st_stid);
1023 1024 1025 1026
		oo->oo_last_closed_stid = NULL;
	}
}

1027 1028 1029 1030 1031 1032
static void nfs4_free_openowner(struct nfs4_openowner *oo)
{
	kfree(oo->oo_owner.so_owner.data);
	kmem_cache_free(openowner_slab, oo);
}

1033
static void release_openowner(struct nfs4_openowner *oo)
1034
{
1035 1036
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
1037
	release_last_closed_stateid(oo);
1038
	nfs4_free_openowner(oo);
1039 1040
}

M
Marc Eshel 已提交
1041 1042 1043 1044 1045 1046 1047 1048
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

1049
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
1050 1051 1052 1053 1054 1055
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]);
}
1056 1057 1058 1059 1060 1061 1062
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
/*
 * 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
1085

A
Andy Adamson 已提交
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
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;
}

/*
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
 * 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)

1112 1113 1114 1115 1116 1117 1118 1119 1120
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]);
}

1121
/*
1122 1123 1124
 * We don't actually need to cache the rpc and session headers, so we
 * can allocate a little less for each slot:
 */
1125
static inline u32 slot_bytes(struct nfsd4_channel_attrs *ca)
1126
{
1127
	u32 size;
1128

1129 1130 1131 1132 1133
	if (ca->maxresp_cached < NFSD_MIN_HDR_SEQ_SZ)
		size = 0;
	else
		size = ca->maxresp_cached - NFSD_MIN_HDR_SEQ_SZ;
	return size + sizeof(struct nfsd4_slot);
1134
}
A
Andy Adamson 已提交
1135

1136 1137
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
1138
 * re-negotiate active sessions and reduce their slot usage to make
1139
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
1140
 */
1141
static u32 nfsd4_get_drc_mem(struct nfsd4_channel_attrs *ca)
A
Andy Adamson 已提交
1142
{
1143 1144
	u32 slotsize = slot_bytes(ca);
	u32 num = ca->maxreqs;
1145
	int avail;
A
Andy Adamson 已提交
1146

1147
	spin_lock(&nfsd_drc_lock);
1148 1149
	avail = min((unsigned long)NFSD_MAX_MEM_PER_SESSION,
		    nfsd_drc_max_mem - nfsd_drc_mem_used);
1150 1151 1152
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
1153

1154 1155
	return num;
}
A
Andy Adamson 已提交
1156

1157
static void nfsd4_put_drc_mem(struct nfsd4_channel_attrs *ca)
1158
{
1159 1160
	int slotsize = slot_bytes(ca);

1161
	spin_lock(&nfsd_drc_lock);
1162
	nfsd_drc_mem_used -= slotsize * ca->maxreqs;
1163
	spin_unlock(&nfsd_drc_lock);
1164
}
A
Andy Adamson 已提交
1165

1166 1167
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fattrs,
					   struct nfsd4_channel_attrs *battrs)
1168
{
1169 1170
	int numslots = fattrs->maxreqs;
	int slotsize = slot_bytes(fattrs);
1171 1172
	struct nfsd4_session *new;
	int mem, i;
1173

1174 1175 1176
	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
1177

1178 1179 1180
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
1181
	/* allocate each struct nfsd4_slot and data cache in one piece */
1182
	for (i = 0; i < numslots; i++) {
1183
		new->se_slots[i] = kzalloc(slotsize, GFP_KERNEL);
1184
		if (!new->se_slots[i])
1185 1186
			goto out_free;
	}
1187 1188 1189 1190

	memcpy(&new->se_fchannel, fattrs, sizeof(struct nfsd4_channel_attrs));
	memcpy(&new->se_bchannel, battrs, sizeof(struct nfsd4_channel_attrs));

1191 1192 1193 1194 1195 1196
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
1197 1198
}

1199 1200 1201 1202 1203
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
1204

1205 1206 1207 1208
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
1209

1210 1211 1212 1213 1214
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
1215
	nfsd4_probe_callback(clp);
1216
	spin_unlock(&clp->cl_lock);
1217
}
A
Andy Adamson 已提交
1218

1219
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
1220 1221
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
1222

1223 1224
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
1225
		return NULL;
1226 1227
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
1228
	conn->cn_flags = flags;
1229 1230 1231
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
1232

1233 1234 1235 1236
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
1237 1238
}

1239
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
1240
{
1241
	struct nfs4_client *clp = ses->se_client;
1242

1243
	spin_lock(&clp->cl_lock);
1244
	__nfsd4_hash_conn(conn, ses);
1245
	spin_unlock(&clp->cl_lock);
1246 1247
}

1248
static int nfsd4_register_conn(struct nfsd4_conn *conn)
1249
{
1250
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
1251
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
1252 1253
}

1254
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
1255
{
1256
	int ret;
A
Andy Adamson 已提交
1257

1258
	nfsd4_hash_conn(conn, ses);
1259 1260 1261 1262
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
1263 1264
	/* We may have gained or lost a callback channel: */
	nfsd4_probe_callback_sync(ses->se_client);
1265
}
A
Andy Adamson 已提交
1266

1267
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
1268 1269 1270
{
	u32 dir = NFS4_CDFC4_FORE;

1271
	if (cses->flags & SESSION4_BACK_CHAN)
1272
		dir |= NFS4_CDFC4_BACK;
1273
	return alloc_conn(rqstp, dir);
1274 1275 1276
}

/* must be called under client_lock */
1277
static void nfsd4_del_conns(struct nfsd4_session *s)
1278
{
1279 1280
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
1281

1282 1283 1284 1285 1286
	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);
1287

1288 1289
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
1290

1291 1292 1293
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
1294
}
A
Andy Adamson 已提交
1295

1296 1297 1298 1299 1300 1301
static void __free_session(struct nfsd4_session *ses)
{
	free_session_slots(ses);
	kfree(ses);
}

1302
static void free_session(struct nfsd4_session *ses)
1303
{
1304
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);
1305 1306

	lockdep_assert_held(&nn->client_lock);
1307
	nfsd4_del_conns(ses);
1308
	nfsd4_put_drc_mem(&ses->se_fchannel);
1309
	__free_session(ses);
1310 1311
}

1312
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
1313 1314
{
	int idx;
1315
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1316

A
Andy Adamson 已提交
1317 1318 1319
	new->se_client = clp;
	gen_sessionid(new);

1320 1321
	INIT_LIST_HEAD(&new->se_conns);

1322
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
1323
	new->se_flags = cses->flags;
1324
	new->se_cb_prog = cses->callback_prog;
1325
	new->se_cb_sec = cses->cb_sec;
1326
	atomic_set(&new->se_ref, 0);
1327
	idx = hash_sessionid(&new->se_sessionid);
1328
	spin_lock(&nn->client_lock);
1329
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
1330
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
1331
	list_add(&new->se_perclnt, &clp->cl_sessions);
1332
	spin_unlock(&clp->cl_lock);
1333
	spin_unlock(&nn->client_lock);
1334

1335
	if (cses->flags & SESSION4_BACK_CHAN) {
1336
		struct sockaddr *sa = svc_addr(rqstp);
1337 1338 1339 1340 1341 1342 1343
		/*
		 * 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:
		 */
1344 1345 1346
		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
1347 1348
}

1349
/* caller must hold client_lock */
M
Marc Eshel 已提交
1350
static struct nfsd4_session *
1351
__find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
1352 1353 1354
{
	struct nfsd4_session *elem;
	int idx;
1355
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
1356 1357 1358 1359

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
1360
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388
static struct nfsd4_session *
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net,
		__be32 *ret)
{
	struct nfsd4_session *session;
	__be32 status = nfserr_badsession;

	session = __find_in_sessionid_hashtbl(sessionid, net);
	if (!session)
		goto out;
	status = nfsd4_get_session_locked(session);
	if (status)
		session = NULL;
out:
	*ret = status;
	return session;
}

1389
/* caller must hold client_lock */
A
Andy Adamson 已提交
1390
static void
M
Marc Eshel 已提交
1391
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
1392 1393
{
	list_del(&ses->se_hash);
1394
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
1395
	list_del(&ses->se_perclnt);
1396
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
1397 1398
}

L
Linus Torvalds 已提交
1399 1400
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1401
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1402
{
1403
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1404
		return 0;
A
Andy Adamson 已提交
1405
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1406
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1407 1408 1409 1410 1411 1412 1413 1414
	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.
 */
1415
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1416 1417 1418
{
	struct nfs4_client *clp;

1419 1420 1421
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1422
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1423 1424 1425
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1426
	}
1427
	clp->cl_name.len = name.len;
1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439
	INIT_LIST_HEAD(&clp->cl_sessions);
	idr_init(&clp->cl_stateids);
	atomic_set(&clp->cl_refcount, 0);
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
	INIT_LIST_HEAD(&clp->cl_callbacks);
	INIT_LIST_HEAD(&clp->cl_revoked);
	spin_lock_init(&clp->cl_lock);
	rpc_init_wait_queue(&clp->cl_cb_waitq, "Backchannel slot table");
L
Linus Torvalds 已提交
1440 1441 1442
	return clp;
}

1443
static void
L
Linus Torvalds 已提交
1444 1445
free_client(struct nfs4_client *clp)
{
1446
	struct nfsd_net __maybe_unused *nn = net_generic(clp->net, nfsd_net_id);
1447 1448

	lockdep_assert_held(&nn->client_lock);
1449 1450 1451 1452 1453
	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);
1454 1455
		WARN_ON_ONCE(atomic_read(&ses->se_ref));
		free_session(ses);
1456
	}
1457
	rpc_destroy_wait_queue(&clp->cl_cb_waitq);
1458
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1459
	kfree(clp->cl_name.data);
M
majianpeng 已提交
1460
	idr_destroy(&clp->cl_stateids);
L
Linus Torvalds 已提交
1461 1462 1463
	kfree(clp);
}

B
Benny Halevy 已提交
1464 1465 1466 1467
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1468 1469
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1470
	list_del(&clp->cl_lru);
1471
	spin_lock(&clp->cl_lock);
1472 1473
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1474
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1475 1476
}

L
Linus Torvalds 已提交
1477
static void
1478
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1479
{
1480
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1481 1482
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1483
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1484 1485

	INIT_LIST_HEAD(&reaplist);
1486
	spin_lock(&state_lock);
1487 1488
	while (!list_empty(&clp->cl_delegations)) {
		dp = list_entry(clp->cl_delegations.next, struct nfs4_delegation, dl_perclnt);
1489 1490
		unhash_delegation_locked(dp);
		list_add(&dp->dl_recall_lru, &reaplist);
L
Linus Torvalds 已提交
1491
	}
1492
	spin_unlock(&state_lock);
L
Linus Torvalds 已提交
1493 1494
	while (!list_empty(&reaplist)) {
		dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
1495
		list_del_init(&dp->dl_recall_lru);
1496
		nfs4_put_stid(&dp->dl_stid);
L
Linus Torvalds 已提交
1497
	}
1498
	while (!list_empty(&clp->cl_revoked)) {
1499
		dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
1500
		list_del_init(&dp->dl_recall_lru);
1501
		nfs4_put_stid(&dp->dl_stid);
1502
	}
1503
	while (!list_empty(&clp->cl_openowners)) {
1504 1505
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1506
	}
1507
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1508 1509
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1510
	list_del(&clp->cl_idhash);
1511
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1512
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1513
	else
1514
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1515
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1516
	unhash_client_locked(clp);
1517 1518
	WARN_ON_ONCE(atomic_read(&clp->cl_refcount));
	free_client(clp);
1519
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1520 1521
}

1522 1523 1524 1525 1526 1527
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1528 1529 1530 1531
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
1532 1533
}

1534 1535
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1536 1537 1538 1539
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1540
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1541
{
1542 1543 1544 1545 1546 1547 1548
	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;
1549
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1550 1551 1552 1553
	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);
1554 1555 1556
	target->cr_gss_mech = source->cr_gss_mech;
	if (source->cr_gss_mech)
		gss_mech_get(source->cr_gss_mech);
1557
	return 0;
L
Linus Torvalds 已提交
1558 1559
}

1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
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);
}

1571
static int same_name(const char *n1, const char *n2)
1572
{
N
NeilBrown 已提交
1573
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1574 1575 1576
}

static int
1577 1578 1579
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1580 1581 1582
}

static int
1583 1584 1585
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
1586 1587
}

1588 1589 1590 1591 1592 1593 1594
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++)
1595
		if (!gid_eq(GROUP_AT(g1, i), GROUP_AT(g2, i)))
1596 1597 1598 1599
			return false;
	return true;
}

1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
/*
 * 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);
}


1616
static bool
1617 1618
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1619
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1620 1621
		|| (!uid_eq(cr1->cr_uid, cr2->cr_uid))
		|| (!gid_eq(cr1->cr_gid, cr2->cr_gid))
1622 1623 1624 1625 1626 1627
		|| !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;
1628
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1629 1630
}

1631 1632 1633 1634 1635
static bool svc_rqst_integrity_protected(struct svc_rqst *rqstp)
{
	struct svc_cred *cr = &rqstp->rq_cred;
	u32 service;

1636 1637
	if (!cr->cr_gss_mech)
		return false;
1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
	service = gss_pseudoflavor_to_service(cr->cr_gss_mech, cr->cr_flavor);
	return service == RPC_GSS_SVC_INTEGRITY ||
	       service == RPC_GSS_SVC_PRIVACY;
}

static bool mach_creds_match(struct nfs4_client *cl, struct svc_rqst *rqstp)
{
	struct svc_cred *cr = &rqstp->rq_cred;

	if (!cl->cl_mach_cred)
		return true;
	if (cl->cl_cred.cr_gss_mech != cr->cr_gss_mech)
		return false;
	if (!svc_rqst_integrity_protected(rqstp))
		return false;
	if (!cr->cr_principal)
		return false;
	return 0 == strcmp(cl->cl_cred.cr_principal, cr->cr_principal);
}

1658
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1659 1660 1661
{
	static u32 current_clientid = 1;

1662
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1663 1664 1665
	clp->cl_clientid.cl_id = current_clientid++; 
}

1666 1667
static void gen_confirm(struct nfs4_client *clp)
{
1668
	__be32 verf[2];
1669
	static u32 i;
L
Linus Torvalds 已提交
1670

1671 1672 1673 1674 1675 1676
	/*
	 * This is opaque to client, so no need to byte-swap. Use
	 * __force to keep sparse happy
	 */
	verf[0] = (__force __be32)get_seconds();
	verf[1] = (__force __be32)i++;
1677
	memcpy(clp->cl_confirm.data, verf, sizeof(clp->cl_confirm.data));
L
Linus Torvalds 已提交
1678 1679
}

1680 1681
static struct nfs4_stid *
find_stateid_locked(struct nfs4_client *cl, stateid_t *t)
1682
{
J
J. Bruce Fields 已提交
1683 1684 1685 1686 1687 1688
	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
1689 1690
}

1691 1692
static struct nfs4_stid *
find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1693 1694
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1695

1696 1697 1698 1699 1700 1701
	spin_lock(&cl->cl_lock);
	s = find_stateid_locked(cl, t);
	if (s != NULL && !(typemask & s->sc_type))
		s = NULL;
	spin_unlock(&cl->cl_lock);
	return s;
1702 1703
}

J
Jeff Layton 已提交
1704
static struct nfs4_client *create_client(struct xdr_netobj name,
1705 1706 1707 1708
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1709
	int ret;
1710
	struct net *net = SVC_NET(rqstp);
1711
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1712 1713 1714 1715 1716

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

1717 1718
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1719
		spin_lock(&nn->client_lock);
1720
		free_client(clp);
1721
		spin_unlock(&nn->client_lock);
1722
		return NULL;
1723
	}
1724
	INIT_WORK(&clp->cl_cb_null.cb_work, nfsd4_run_cb_null);
B
Benny Halevy 已提交
1725
	clp->cl_time = get_seconds();
1726 1727 1728 1729
	clear_bit(0, &clp->cl_cb_slot_busy);
	copy_verf(clp, verf);
	rpc_copy_addr((struct sockaddr *) &clp->cl_addr, sa);
	gen_confirm(clp);
1730
	clp->cl_cb_session = NULL;
1731
	clp->net = net;
1732 1733 1734
	return clp;
}

1735
static void
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
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
1777 1778
{
	unsigned int idhashval;
1779
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1780

1781
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1782
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1783
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1784
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1785
	renew_client(clp);
L
Linus Torvalds 已提交
1786 1787
}

1788
static void
L
Linus Torvalds 已提交
1789 1790 1791
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1792
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1793 1794

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1795
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1796
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1797
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1798
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1799 1800 1801 1802
	renew_client(clp);
}

static struct nfs4_client *
J
J. Bruce Fields 已提交
1803
find_client_in_id_table(struct list_head *tbl, clientid_t *clid, bool sessions)
L
Linus Torvalds 已提交
1804 1805 1806 1807
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

J
J. Bruce Fields 已提交
1808
	list_for_each_entry(clp, &tbl[idhashval], cl_idhash) {
1809
		if (same_clid(&clp->cl_clientid, clid)) {
1810 1811
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1812
			renew_client(clp);
L
Linus Torvalds 已提交
1813
			return clp;
1814
		}
L
Linus Torvalds 已提交
1815 1816 1817 1818
	}
	return NULL;
}

J
J. Bruce Fields 已提交
1819 1820 1821 1822 1823 1824 1825 1826
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
1827
static struct nfs4_client *
1828
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1829
{
J
J. Bruce Fields 已提交
1830
	struct list_head *tbl = nn->unconf_id_hashtbl;
L
Linus Torvalds 已提交
1831

J
J. Bruce Fields 已提交
1832
	return find_client_in_id_table(tbl, clid, sessions);
L
Linus Torvalds 已提交
1833 1834
}

1835
static bool clp_used_exchangeid(struct nfs4_client *clp)
1836
{
1837
	return clp->cl_exchange_flags != 0;
1838
} 
1839

1840
static struct nfs4_client *
1841
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1842
{
1843
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1844 1845 1846
}

static struct nfs4_client *
1847
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1848
{
1849
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1850 1851
}

1852
static void
1853
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1854
{
1855
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1856 1857
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
	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
1868 1869
		goto out_err;

1870
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1871
					    se->se_callback_addr_len,
1872 1873
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1874

1875
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1876
		goto out_err;
1877

1878 1879
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1880

1881 1882
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1883
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1884 1885
	return;
out_err:
1886 1887
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1888
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1889 1890 1891 1892 1893 1894
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1895
/*
1896
 * Cache a reply. nfsd4_check_resp_size() has bounded the cache size.
1897
 */
1898
static void
1899 1900
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1901
	struct xdr_buf *buf = resp->xdr.buf;
1902 1903
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1904

1905
	dprintk("--> %s slot %p\n", __func__, slot);
1906

1907 1908
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1909

1910
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1911
	if (nfsd4_not_cached(resp)) {
1912
		slot->sl_datalen = 0;
1913
		return;
1914
	}
1915 1916 1917
	base = resp->cstate.data_offset;
	slot->sl_datalen = buf->len - base;
	if (read_bytes_from_xdr_buf(buf, base, slot->sl_data, slot->sl_datalen))
1918 1919
		WARN("%s: sessions DRC could not cache compound\n", __func__);
	return;
1920 1921 1922
}

/*
1923 1924 1925 1926
 * 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.
1927 1928
 *
 */
1929 1930 1931
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1932
{
1933 1934
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1935

1936 1937 1938
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1939

1940
	/* Return nfserr_retry_uncached_rep in next operation. */
1941
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1942 1943 1944
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1945
	}
1946
	return op->status;
1947 1948 1949
}

/*
1950 1951
 * The sequence operation is not cached because we can use the slot and
 * session values.
1952
 */
1953
static __be32
1954 1955
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1956
{
1957
	struct nfsd4_slot *slot = resp->cstate.slot;
1958 1959
	struct xdr_stream *xdr = &resp->xdr;
	__be32 *p;
1960 1961
	__be32 status;

1962
	dprintk("--> %s slot %p\n", __func__, slot);
1963

1964
	status = nfsd4_enc_sequence_replay(resp->rqstp->rq_argp, resp);
1965
	if (status)
1966
		return status;
1967

1968 1969 1970 1971 1972 1973 1974
	p = xdr_reserve_space(xdr, slot->sl_datalen);
	if (!p) {
		WARN_ON_ONCE(1);
		return nfserr_serverfault;
	}
	xdr_encode_opaque_fixed(p, slot->sl_data, slot->sl_datalen);
	xdr_commit_encode(xdr);
1975

1976
	resp->opcnt = slot->sl_opcnt;
1977
	return slot->sl_status;
1978 1979
}

A
Andy Adamson 已提交
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
/*
 * 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;
}

1996 1997 1998 1999 2000 2001 2002 2003
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.
	 */
2004 2005 2006
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
2007 2008
}

A
Andy Adamson 已提交
2009 2010 2011 2012 2013
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
2014
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
2015
	__be32 status;
2016
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
2017
	nfs4_verifier		verf = exid->verifier;
2018
	struct sockaddr		*sa = svc_addr(rqstp);
2019
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
2020
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
2021

2022
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
2023
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
2024
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
2025
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
2026
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
2027

2028
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
2029 2030 2031
		return nfserr_inval;

	switch (exid->spa_how) {
2032 2033 2034
	case SP4_MACH_CRED:
		if (!svc_rqst_integrity_protected(rqstp))
			return nfserr_inval;
A
Andy Adamson 已提交
2035 2036
	case SP4_NONE:
		break;
2037 2038
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
2039
	case SP4_SSV:
2040
		return nfserr_encr_alg_unsupp;
A
Andy Adamson 已提交
2041 2042
	}

2043
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
2044
	nfs4_lock_state();
2045
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
2046
	if (conf) {
2047 2048 2049
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

2050 2051
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
2052
				status = nfserr_inval;
2053 2054
				goto out;
			}
2055 2056 2057 2058
			if (!mach_creds_match(conf, rqstp)) {
				status = nfserr_wrong_cred;
				goto out;
			}
2059
			if (!creds_match) { /* case 9 */
2060
				status = nfserr_perm;
2061 2062 2063
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
2064 2065 2066
				status = nfserr_not_same;
				goto out;
			}
2067 2068 2069 2070
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
2071
		}
2072
		if (!creds_match) { /* case 3 */
2073 2074
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
2075 2076 2077 2078 2079
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
2080
		if (verfs_match) { /* case 2 */
2081
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
2082 2083 2084 2085 2086
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
2087 2088
	}

2089
	if (update) { /* case 7 */
2090 2091
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
2092 2093
	}

2094
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
2095
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
2096 2097
		expire_client(unconf);

2098
	/* case 1 (normal case) */
A
Andy Adamson 已提交
2099
out_new:
J
Jeff Layton 已提交
2100
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
2101
	if (new == NULL) {
2102
		status = nfserr_jukebox;
A
Andy Adamson 已提交
2103 2104
		goto out;
	}
2105
	new->cl_minorversion = cstate->minorversion;
2106
	new->cl_mach_cred = (exid->spa_how == SP4_MACH_CRED);
A
Andy Adamson 已提交
2107

2108
	gen_clid(new, nn);
2109
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
2110 2111 2112 2113
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

2114
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
2115 2116 2117
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
2118
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
2119 2120 2121 2122 2123
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
2124 2125
}

J
J. Bruce Fields 已提交
2126
static __be32
2127
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
2128
{
2129 2130
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
2131 2132

	/* The slot is in use, and no response has been sent. */
2133 2134
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
2135 2136 2137 2138
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
2139
	/* Note unsigned 32-bit arithmetic handles wraparound: */
2140
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
2141
		return nfs_ok;
2142
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
2143 2144 2145 2146
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

2147 2148 2149 2150 2151 2152 2153
/*
 * 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
2154
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167
{
	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;
}

2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184
#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))

2185
static __be32 check_forechannel_attrs(struct nfsd4_channel_attrs *ca, struct nfsd_net *nn)
2186
{
2187 2188
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;

2189 2190 2191 2192
	if (ca->maxreq_sz < NFSD_MIN_REQ_HDR_SEQ_SZ)
		return nfserr_toosmall;
	if (ca->maxresp_sz < NFSD_MIN_RESP_HDR_SEQ_SZ)
		return nfserr_toosmall;
2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211
	ca->headerpadsz = 0;
	ca->maxreq_sz = min_t(u32, ca->maxreq_sz, maxrpc);
	ca->maxresp_sz = min_t(u32, ca->maxresp_sz, maxrpc);
	ca->maxops = min_t(u32, ca->maxops, NFSD_MAX_OPS_PER_COMPOUND);
	ca->maxresp_cached = min_t(u32, ca->maxresp_cached,
			NFSD_SLOT_CACHE_SIZE + NFSD_MIN_HDR_SEQ_SZ);
	ca->maxreqs = min_t(u32, ca->maxreqs, NFSD_MAX_SLOTS_PER_SESSION);
	/*
	 * 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.  Clients that
	 * request larger slots than they need will get poor results:
	 */
	ca->maxreqs = nfsd4_get_drc_mem(ca);
	if (!ca->maxreqs)
		return nfserr_jukebox;

2212
	return nfs_ok;
2213 2214
}

2215 2216 2217 2218 2219
#define NFSD_CB_MAX_REQ_SZ	((NFS4_enc_cb_recall_sz + \
				 RPC_MAX_HEADER_WITH_AUTH) * sizeof(__be32))
#define NFSD_CB_MAX_RESP_SZ	((NFS4_dec_cb_recall_sz + \
				 RPC_MAX_REPHEADER_WITH_AUTH) * sizeof(__be32))

2220
static __be32 check_backchannel_attrs(struct nfsd4_channel_attrs *ca)
2221
{
2222 2223 2224 2225 2226 2227 2228 2229
	ca->headerpadsz = 0;

	/*
	 * These RPC_MAX_HEADER macros are overkill, especially since we
	 * don't even do gss on the backchannel yet.  But this is still
	 * less than 1k.  Tighten up this estimate in the unlikely event
	 * it turns out to be a problem for some client:
	 */
2230
	if (ca->maxreq_sz < NFSD_CB_MAX_REQ_SZ)
2231
		return nfserr_toosmall;
2232
	if (ca->maxresp_sz < NFSD_CB_MAX_RESP_SZ)
2233 2234 2235 2236 2237 2238
		return nfserr_toosmall;
	ca->maxresp_cached = 0;
	if (ca->maxops < 2)
		return nfserr_toosmall;

	return nfs_ok;
2239 2240
}

2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258
static __be32 nfsd4_check_cb_sec(struct nfsd4_cb_sec *cbs)
{
	switch (cbs->flavor) {
	case RPC_AUTH_NULL:
	case RPC_AUTH_UNIX:
		return nfs_ok;
	default:
		/*
		 * GSS case: the spec doesn't allow us to return this
		 * error.  But it also doesn't allow us not to support
		 * GSS.
		 * I'd rather this fail hard than return some error the
		 * client might think it can already handle:
		 */
		return nfserr_encr_alg_unsupp;
	}
}

A
Andy Adamson 已提交
2259 2260 2261 2262 2263
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
2264
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
2265
	struct nfs4_client *conf, *unconf;
2266
	struct nfsd4_session *new;
2267
	struct nfsd4_conn *conn;
2268
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
2269
	__be32 status = 0;
2270
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
2271

2272 2273
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
2274 2275 2276
	status = nfsd4_check_cb_sec(&cr_ses->cb_sec);
	if (status)
		return status;
2277
	status = check_forechannel_attrs(&cr_ses->fore_channel, nn);
2278 2279 2280
	if (status)
		return status;
	status = check_backchannel_attrs(&cr_ses->back_channel);
2281
	if (status)
2282
		goto out_release_drc_mem;
2283
	status = nfserr_jukebox;
2284
	new = alloc_session(&cr_ses->fore_channel, &cr_ses->back_channel);
2285 2286
	if (!new)
		goto out_release_drc_mem;
2287 2288 2289
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
2290

A
Andy Adamson 已提交
2291
	nfs4_lock_state();
2292
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
2293
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
2294
	WARN_ON_ONCE(conf && unconf);
A
Andy Adamson 已提交
2295 2296

	if (conf) {
2297 2298 2299
		status = nfserr_wrong_cred;
		if (!mach_creds_match(conf, rqstp))
			goto out_free_conn;
2300 2301
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
2302
		if (status == nfserr_replay_cache) {
2303
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
2304
			goto out_free_conn;
2305
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
2306
			status = nfserr_seq_misordered;
2307
			goto out_free_conn;
A
Andy Adamson 已提交
2308 2309
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
2310
		struct nfs4_client *old;
A
Andy Adamson 已提交
2311
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
2312
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
2313
			status = nfserr_clid_inuse;
2314
			goto out_free_conn;
A
Andy Adamson 已提交
2315
		}
2316 2317 2318
		status = nfserr_wrong_cred;
		if (!mach_creds_match(unconf, rqstp))
			goto out_free_conn;
2319 2320
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
2321 2322
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
2323
			status = nfserr_seq_misordered;
2324
			goto out_free_conn;
A
Andy Adamson 已提交
2325
		}
2326
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
2327 2328 2329 2330
		if (old) {
			status = mark_client_expired(old);
			if (status)
				goto out_free_conn;
J
J. Bruce Fields 已提交
2331
			expire_client(old);
2332
		}
J
J. Bruce Fields 已提交
2333
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
2334 2335 2336
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
2337
		goto out_free_conn;
A
Andy Adamson 已提交
2338
	}
2339
	status = nfs_ok;
2340 2341 2342 2343 2344 2345
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

2346 2347 2348
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

2349
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
2350
	       NFS4_MAX_SESSIONID_LEN);
2351
	cs_slot->sl_seqid++;
2352
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
2353

2354 2355
	/* 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
2356 2357
	nfs4_unlock_state();
	return status;
2358
out_free_conn:
2359
	nfs4_unlock_state();
2360 2361 2362
	free_conn(conn);
out_free_session:
	__free_session(new);
2363 2364
out_release_drc_mem:
	nfsd4_put_drc_mem(&cr_ses->fore_channel);
J
J. Bruce Fields 已提交
2365
	return status;
A
Andy Adamson 已提交
2366 2367
}

2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
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;
}

2382 2383 2384
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
2385
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2386
	__be32 status;
2387

2388 2389 2390
	status = nfsd4_check_cb_sec(&bc->bc_cb_sec);
	if (status)
		return status;
2391
	spin_lock(&nn->client_lock);
2392 2393
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
2394
	spin_unlock(&nn->client_lock);
2395 2396 2397 2398 2399 2400

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

2401 2402 2403 2404 2405
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
2406
	struct nfsd4_conn *conn;
2407
	struct nfsd4_session *session;
2408 2409
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
2410 2411 2412

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
2413
	nfs4_lock_state();
2414
	spin_lock(&nn->client_lock);
2415
	session = find_in_sessionid_hashtbl(&bcts->sessionid, net, &status);
2416
	spin_unlock(&nn->client_lock);
2417
	if (!session)
2418
		goto out_no_session;
2419 2420 2421
	status = nfserr_wrong_cred;
	if (!mach_creds_match(session->se_client, rqstp))
		goto out;
2422
	status = nfsd4_map_bcts_dir(&bcts->dir);
2423
	if (status)
2424
		goto out;
2425
	conn = alloc_conn(rqstp, bcts->dir);
2426
	status = nfserr_jukebox;
2427
	if (!conn)
2428 2429 2430 2431
		goto out;
	nfsd4_init_conn(rqstp, conn, session);
	status = nfs_ok;
out:
2432 2433
	nfsd4_put_session(session);
out_no_session:
2434 2435
	nfs4_unlock_state();
	return status;
2436 2437
}

2438 2439 2440 2441 2442 2443 2444
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
2445 2446 2447 2448 2449
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
2450
	struct nfsd4_session *ses;
2451
	__be32 status;
2452
	int ref_held_by_me = 0;
2453 2454
	struct net *net = SVC_NET(r);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
B
Benny Halevy 已提交
2455

2456 2457
	nfs4_lock_state();
	status = nfserr_not_only_op;
2458
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
2459
		if (!nfsd4_last_compound_op(r))
2460
			goto out;
2461
		ref_held_by_me++;
2462
	}
B
Benny Halevy 已提交
2463
	dump_sessionid(__func__, &sessionid->sessionid);
2464
	spin_lock(&nn->client_lock);
2465
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, net, &status);
2466 2467
	if (!ses)
		goto out_client_lock;
2468 2469
	status = nfserr_wrong_cred;
	if (!mach_creds_match(ses->se_client, r))
2470
		goto out_put_session;
2471
	status = mark_session_dead_locked(ses, 1 + ref_held_by_me);
2472
	if (status)
2473
		goto out_put_session;
B
Benny Halevy 已提交
2474
	unhash_session(ses);
2475
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2476

2477
	nfsd4_probe_callback_sync(ses->se_client);
2478

2479
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2480
	status = nfs_ok;
2481
out_put_session:
2482
	nfsd4_put_session_locked(ses);
2483 2484
out_client_lock:
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2485
out:
2486
	nfs4_unlock_state();
B
Benny Halevy 已提交
2487
	return status;
A
Andy Adamson 已提交
2488 2489
}

2490
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
2491 2492 2493 2494
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
2495
		if (c->cn_xprt == xpt) {
2496 2497 2498 2499 2500 2501
			return c;
		}
	}
	return NULL;
}

2502
static __be32 nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
2503 2504
{
	struct nfs4_client *clp = ses->se_client;
2505
	struct nfsd4_conn *c;
2506
	__be32 status = nfs_ok;
2507
	int ret;
2508 2509

	spin_lock(&clp->cl_lock);
2510
	c = __nfsd4_find_conn(new->cn_xprt, ses);
2511 2512 2513 2514 2515
	if (c)
		goto out_free;
	status = nfserr_conn_not_bound_to_session;
	if (clp->cl_mach_cred)
		goto out_free;
2516 2517
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
2518 2519 2520 2521
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2522 2523 2524 2525 2526
	return nfs_ok;
out_free:
	spin_unlock(&clp->cl_lock);
	free_conn(new);
	return status;
2527 2528
}

2529 2530 2531 2532 2533 2534 2535
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
2536 2537 2538 2539 2540 2541 2542 2543
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
2544
__be32
B
Benny Halevy 已提交
2545
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2546 2547 2548
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2549
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
2550
	struct xdr_stream *xdr = &resp->xdr;
B
Benny Halevy 已提交
2551
	struct nfsd4_session *session;
2552
	struct nfs4_client *clp;
B
Benny Halevy 已提交
2553
	struct nfsd4_slot *slot;
2554
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2555
	__be32 status;
2556
	int buflen;
2557 2558
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
B
Benny Halevy 已提交
2559

2560 2561 2562
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2563 2564 2565 2566 2567 2568 2569 2570
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2571
	spin_lock(&nn->client_lock);
2572
	session = find_in_sessionid_hashtbl(&seq->sessionid, net, &status);
B
Benny Halevy 已提交
2573
	if (!session)
2574 2575
		goto out_no_session;
	clp = session->se_client;
B
Benny Halevy 已提交
2576

2577 2578
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
2579
		goto out_put_session;
2580

M
Mi Jinlong 已提交
2581 2582
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
2583
		goto out_put_session;
M
Mi Jinlong 已提交
2584

B
Benny Halevy 已提交
2585
	status = nfserr_badslot;
2586
	if (seq->slotid >= session->se_fchannel.maxreqs)
2587
		goto out_put_session;
B
Benny Halevy 已提交
2588

2589
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2590 2591
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2592 2593 2594 2595 2596
	/* 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;

2597 2598
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2599
	if (status == nfserr_replay_cache) {
2600 2601
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
2602
			goto out_put_session;
B
Benny Halevy 已提交
2603 2604
		cstate->slot = slot;
		cstate->session = session;
2605
		cstate->clp = clp;
A
Andy Adamson 已提交
2606
		/* Return the cached reply status and set cstate->status
2607
		 * for nfsd4_proc_compound processing */
2608
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2609
		cstate->status = nfserr_replay_cache;
2610
		goto out;
B
Benny Halevy 已提交
2611 2612
	}
	if (status)
2613
		goto out_put_session;
B
Benny Halevy 已提交
2614

2615
	status = nfsd4_sequence_check_conn(conn, session);
2616
	conn = NULL;
2617 2618
	if (status)
		goto out_put_session;
2619

2620 2621 2622 2623 2624
	buflen = (seq->cachethis) ?
			session->se_fchannel.maxresp_cached :
			session->se_fchannel.maxresp_sz;
	status = (seq->cachethis) ? nfserr_rep_too_big_to_cache :
				    nfserr_rep_too_big;
2625
	if (xdr_restrict_buflen(xdr, buflen - rqstp->rq_auth_slack))
2626
		goto out_put_session;
2627
	svc_reserve(rqstp, buflen);
2628 2629

	status = nfs_ok;
B
Benny Halevy 已提交
2630 2631
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2632
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2633 2634
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2635 2636
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2637 2638 2639

	cstate->slot = slot;
	cstate->session = session;
2640
	cstate->clp = clp;
B
Benny Halevy 已提交
2641 2642

out:
2643 2644 2645 2646 2647 2648 2649 2650 2651
	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;
2652
	}
2653 2654
	if (!list_empty(&clp->cl_revoked))
		seq->status_flags |= SEQ4_STATUS_RECALLABLE_STATE_REVOKED;
2655
out_no_session:
2656 2657
	if (conn)
		free_conn(conn);
2658
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2659
	return status;
2660
out_put_session:
2661
	nfsd4_put_session_locked(session);
2662
	goto out_no_session;
A
Andy Adamson 已提交
2663 2664
}

2665 2666 2667 2668 2669 2670 2671 2672 2673 2674
void
nfsd4_sequence_done(struct nfsd4_compoundres *resp)
{
	struct nfsd4_compound_state *cs = &resp->cstate;

	if (nfsd4_has_session(cs)) {
		if (cs->status != nfserr_replay_cache) {
			nfsd4_store_cache_entry(resp);
			cs->slot->sl_flags &= ~NFSD4_SLOT_INUSE;
		}
2675
		/* Drop session reference that was taken in nfsd4_sequence() */
2676
		nfsd4_put_session(cs->session);
2677 2678
	} else if (cs->clp)
		put_client_renew(cs->clp);
2679 2680
}

2681 2682 2683 2684
__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
2685
	__be32 status = 0;
2686
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2687 2688

	nfs4_lock_state();
2689
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2690
	conf = find_confirmed_client(&dc->clientid, true, nn);
2691
	WARN_ON_ONCE(conf && unconf);
2692 2693 2694 2695

	if (conf) {
		clp = conf;

2696
		if (client_has_state(conf)) {
2697 2698 2699 2700 2701 2702 2703 2704 2705
			status = nfserr_clientid_busy;
			goto out;
		}
	} else if (unconf)
		clp = unconf;
	else {
		status = nfserr_stale_clientid;
		goto out;
	}
2706 2707 2708 2709
	if (!mach_creds_match(clp, rqstp)) {
		status = nfserr_wrong_cred;
		goto out;
	}
2710 2711 2712 2713 2714 2715
	expire_client(clp);
out:
	nfs4_unlock_state();
	return status;
}

2716 2717 2718
__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
2719
	__be32 status = 0;
2720

2721 2722 2723 2724 2725 2726 2727 2728 2729
	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;
	}
2730

2731
	nfs4_lock_state();
2732
	status = nfserr_complete_already;
2733 2734
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2735 2736 2737 2738
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2739 2740 2741 2742 2743 2744 2745
		/*
		 * 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.
		 */
2746 2747 2748
		goto out;

	status = nfs_ok;
2749
	nfsd4_client_record_create(cstate->session->se_client);
2750
out:
2751
	nfs4_unlock_state();
2752
	return status;
2753 2754
}

2755
__be32
2756 2757
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2758
{
2759
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2760
	nfs4_verifier		clverifier = setclid->se_verf;
2761
	struct nfs4_client	*conf, *unconf, *new;
2762
	__be32 			status;
2763 2764
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2765
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2766
	nfs4_lock_state();
2767
	conf = find_confirmed_client_by_name(&clname, nn);
2768
	if (conf) {
2769
		/* case 0: */
L
Linus Torvalds 已提交
2770
		status = nfserr_clid_inuse;
2771 2772
		if (clp_used_exchangeid(conf))
			goto out;
2773
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2774 2775 2776 2777 2778
			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
2779 2780 2781
			goto out;
		}
	}
2782
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2783 2784
	if (unconf)
		expire_client(unconf);
2785
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2786
	new = create_client(clname, rqstp, &clverifier);
2787 2788
	if (new == NULL)
		goto out;
2789
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2790
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2791
		copy_clid(new, conf);
2792
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2793
		gen_clid(new, nn);
2794
	new->cl_minorversion = 0;
2795
	gen_callback(new, setclid, rqstp);
2796
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2797 2798 2799 2800 2801 2802 2803 2804 2805 2806
	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;
}


2807
__be32
2808 2809 2810
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2811
{
2812
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2813 2814
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2815
	__be32 status;
2816
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2817

2818
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2819 2820
		return nfserr_stale_clientid;
	nfs4_lock_state();
2821

2822
	conf = find_confirmed_client(clid, false, nn);
2823
	unconf = find_unconfirmed_client(clid, false, nn);
2824
	/*
2825 2826 2827 2828
	 * 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.
2829
	 */
2830 2831 2832 2833 2834
	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;
2835
	/* cases below refer to rfc 3530 section 14.2.34: */
2836 2837
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2838
			status = nfs_ok;
2839 2840 2841 2842 2843 2844
		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 */
2845 2846 2847
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2848
	} else { /* case 3: normal case; new or rebooted client */
2849
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2850 2851 2852 2853
		if (conf) {
			status = mark_client_expired(conf);
			if (status)
				goto out;
2854
			expire_client(conf);
2855
		}
2856
		move_to_confirmed(unconf);
2857
		nfsd4_probe_callback(unconf);
2858
	}
L
Linus Torvalds 已提交
2859 2860 2861 2862 2863
out:
	nfs4_unlock_state();
	return status;
}

2864 2865 2866 2867 2868
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2869
/* OPEN Share state helper functions */
2870
static void nfsd4_init_file(struct nfs4_file *fp, struct knfsd_fh *fh)
L
Linus Torvalds 已提交
2871
{
2872
	unsigned int hashval = file_hashval(fh);
L
Linus Torvalds 已提交
2873

2874 2875
	lockdep_assert_held(&state_lock);

2876
	atomic_set(&fp->fi_ref, 1);
2877
	spin_lock_init(&fp->fi_lock);
2878 2879
	INIT_LIST_HEAD(&fp->fi_stateids);
	INIT_LIST_HEAD(&fp->fi_delegations);
2880
	fh_copy_shallow(&fp->fi_fhandle, fh);
2881 2882
	fp->fi_had_conflict = false;
	fp->fi_lease = NULL;
2883
	fp->fi_share_deny = 0;
2884 2885
	memset(fp->fi_fds, 0, sizeof(fp->fi_fds));
	memset(fp->fi_access, 0, sizeof(fp->fi_access));
2886
	hlist_add_head(&fp->fi_hash, &file_hashtbl[hashval]);
L
Linus Torvalds 已提交
2887 2888
}

2889
void
L
Linus Torvalds 已提交
2890 2891
nfsd4_free_slabs(void)
{
C
Christoph Hellwig 已提交
2892 2893 2894 2895 2896
	kmem_cache_destroy(openowner_slab);
	kmem_cache_destroy(lockowner_slab);
	kmem_cache_destroy(file_slab);
	kmem_cache_destroy(stateid_slab);
	kmem_cache_destroy(deleg_slab);
N
NeilBrown 已提交
2897
}
L
Linus Torvalds 已提交
2898

2899
int
N
NeilBrown 已提交
2900 2901
nfsd4_init_slabs(void)
{
2902 2903 2904
	openowner_slab = kmem_cache_create("nfsd4_openowners",
			sizeof(struct nfs4_openowner), 0, 0, NULL);
	if (openowner_slab == NULL)
C
Christoph Hellwig 已提交
2905
		goto out;
2906
	lockowner_slab = kmem_cache_create("nfsd4_lockowners",
2907
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2908
	if (lockowner_slab == NULL)
C
Christoph Hellwig 已提交
2909
		goto out_free_openowner_slab;
N
NeilBrown 已提交
2910
	file_slab = kmem_cache_create("nfsd4_files",
2911
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2912
	if (file_slab == NULL)
C
Christoph Hellwig 已提交
2913
		goto out_free_lockowner_slab;
N
NeilBrown 已提交
2914
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2915
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2916
	if (stateid_slab == NULL)
C
Christoph Hellwig 已提交
2917
		goto out_free_file_slab;
N
NeilBrown 已提交
2918
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2919
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2920
	if (deleg_slab == NULL)
C
Christoph Hellwig 已提交
2921
		goto out_free_stateid_slab;
N
NeilBrown 已提交
2922
	return 0;
C
Christoph Hellwig 已提交
2923 2924 2925 2926 2927 2928 2929 2930 2931 2932

out_free_stateid_slab:
	kmem_cache_destroy(stateid_slab);
out_free_file_slab:
	kmem_cache_destroy(file_slab);
out_free_lockowner_slab:
	kmem_cache_destroy(lockowner_slab);
out_free_openowner_slab:
	kmem_cache_destroy(openowner_slab);
out:
N
NeilBrown 已提交
2933 2934
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2935 2936
}

2937
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2938
{
2939 2940 2941
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2942 2943
}

2944
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2945
{
L
Linus Torvalds 已提交
2946 2947
	struct nfs4_stateowner *sop;

2948
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2949 2950 2951 2952 2953
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2954
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2955
		return NULL;
2956 2957 2958
	}
	sop->so_owner.len = owner->len;

2959
	INIT_LIST_HEAD(&sop->so_stateids);
2960 2961 2962 2963 2964
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2965
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2966
{
2967 2968 2969
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2970
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2971 2972
}

2973
static struct nfs4_openowner *
2974
alloc_init_open_stateowner(unsigned int strhashval, struct nfsd4_open *open,
2975 2976
			   struct nfsd4_compound_state *cstate)
{
2977
	struct nfs4_client *clp = cstate->clp;
2978
	struct nfs4_openowner *oo;
2979

2980 2981
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2982
		return NULL;
2983 2984
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2985
	oo->oo_flags = NFS4_OO_NEW;
2986 2987
	if (nfsd4_has_session(cstate))
		oo->oo_flags |= NFS4_OO_CONFIRMED;
2988
	oo->oo_time = 0;
2989
	oo->oo_last_closed_stid = NULL;
2990 2991 2992
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2993 2994
}

2995
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2996
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2997

J
J. Bruce Fields 已提交
2998
	stp->st_stid.sc_type = NFS4_OPEN_STID;
2999
	INIT_LIST_HEAD(&stp->st_locks);
3000
	stp->st_stateowner = &oo->oo_owner;
3001
	get_nfs4_file(fp);
3002
	stp->st_stid.sc_file = fp;
L
Linus Torvalds 已提交
3003 3004
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
3005
	stp->st_openstp = NULL;
3006 3007
	spin_lock(&oo->oo_owner.so_client->cl_lock);
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
3008 3009 3010
	spin_lock(&fp->fi_lock);
	list_add(&stp->st_perfile, &fp->fi_stateids);
	spin_unlock(&fp->fi_lock);
3011
	spin_unlock(&oo->oo_owner.so_client->cl_lock);
L
Linus Torvalds 已提交
3012 3013
}

3014
static void
3015
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
3016
{
3017 3018
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

3019
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
3020

3021
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
3022
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
3023 3024 3025
}

static int
3026 3027 3028 3029 3030 3031
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
3032 3033
}

3034
static struct nfs4_openowner *
3035 3036
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3037
{
3038 3039
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
3040
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
3041

3042
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
3043 3044
		if (!so->so_is_open_owner)
			continue;
3045 3046
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
3047 3048 3049
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
3050 3051 3052
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
3053 3054 3055 3056 3057 3058
	}
	return NULL;
}

/* search file_hashtbl[] for file */
static struct nfs4_file *
3059
find_file_locked(struct knfsd_fh *fh)
L
Linus Torvalds 已提交
3060
{
3061
	unsigned int hashval = file_hashval(fh);
L
Linus Torvalds 已提交
3062 3063
	struct nfs4_file *fp;

3064 3065
	lockdep_assert_held(&state_lock);

3066
	hlist_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
3067
		if (nfsd_fh_match(&fp->fi_fhandle, fh)) {
3068
			get_nfs4_file(fp);
L
Linus Torvalds 已提交
3069
			return fp;
3070
		}
L
Linus Torvalds 已提交
3071 3072 3073 3074
	}
	return NULL;
}

3075
static struct nfs4_file *
3076
find_file(struct knfsd_fh *fh)
3077 3078 3079 3080
{
	struct nfs4_file *fp;

	spin_lock(&state_lock);
3081
	fp = find_file_locked(fh);
3082 3083 3084 3085 3086
	spin_unlock(&state_lock);
	return fp;
}

static struct nfs4_file *
3087
find_or_add_file(struct nfs4_file *new, struct knfsd_fh *fh)
3088 3089 3090 3091
{
	struct nfs4_file *fp;

	spin_lock(&state_lock);
3092
	fp = find_file_locked(fh);
3093
	if (fp == NULL) {
3094
		nfsd4_init_file(new, fh);
3095 3096 3097 3098 3099 3100 3101
		fp = new;
	}
	spin_unlock(&state_lock);

	return fp;
}

L
Linus Torvalds 已提交
3102 3103 3104 3105
/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
3106
static __be32
L
Linus Torvalds 已提交
3107 3108 3109
nfs4_share_conflict(struct svc_fh *current_fh, unsigned int deny_type)
{
	struct nfs4_file *fp;
3110
	__be32 ret = nfs_ok;
L
Linus Torvalds 已提交
3111

3112
	fp = find_file(&current_fh->fh_handle);
3113
	if (!fp)
3114 3115
		return ret;
	/* Check for conflicting share reservations */
3116
	spin_lock(&fp->fi_lock);
3117 3118
	if (fp->fi_share_deny & deny_type)
		ret = nfserr_locked;
3119
	spin_unlock(&fp->fi_lock);
3120 3121
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
3122 3123
}

3124
void nfsd4_prepare_cb_recall(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
3125
{
3126 3127
	struct nfsd_net *nn = net_generic(dp->dl_stid.sc_client->net,
					  nfsd_net_id);
3128

3129
	block_delegations(&dp->dl_stid.sc_file->fi_fhandle);
3130

3131
	/*
3132 3133 3134
	 * We can't do this in nfsd_break_deleg_cb because it is
	 * already holding inode->i_lock.
	 *
3135 3136 3137
	 * If the dl_time != 0, then we know that it has already been
	 * queued for a lease break. Don't queue it again.
	 */
3138
	spin_lock(&state_lock);
3139 3140
	if (dp->dl_time == 0) {
		dp->dl_time = get_seconds();
3141
		list_add_tail(&dp->dl_recall_lru, &nn->del_recall_lru);
3142
	}
3143 3144
	spin_unlock(&state_lock);
}
L
Linus Torvalds 已提交
3145

3146 3147 3148 3149 3150 3151 3152 3153 3154
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
{
	/*
	 * 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.
	 */
3155
	atomic_inc(&dp->dl_stid.sc_count);
3156 3157 3158
	nfsd4_cb_recall(dp);
}

3159
/* Called from break_lease() with i_lock held. */
3160 3161
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
3162 3163
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
3164

3165 3166 3167 3168 3169 3170 3171 3172
	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;
	}
3173 3174
	/*
	 * We don't want the locks code to timeout the lease for us;
3175
	 * we'll remove it ourself if a delegation isn't returned
3176
	 * in time:
3177 3178
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
3179

3180
	spin_lock(&fp->fi_lock);
3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192
	fp->fi_had_conflict = true;
	/*
	 * If there are no delegations on the list, then we can't count on this
	 * lease ever being cleaned up. Set the fl_break_time to jiffies so that
	 * time_out_leases will do it ASAP. The fact that fi_had_conflict is now
	 * true should keep any new delegations from being hashed.
	 */
	if (list_empty(&fp->fi_delegations))
		fl->fl_break_time = jiffies;
	else
		list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
			nfsd_break_one_deleg(dp);
3193
	spin_unlock(&fp->fi_lock);
L
Linus Torvalds 已提交
3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
}

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

3205
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
3206 3207
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
3208 3209
};

3210 3211 3212 3213 3214 3215 3216 3217 3218 3219
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
3220

3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252
static __be32 lookup_clientid(clientid_t *clid,
		struct nfsd4_compound_state *cstate,
		struct nfsd_net *nn)
{
	struct nfs4_client *found;

	if (cstate->clp) {
		found = cstate->clp;
		if (!same_clid(&found->cl_clientid, clid))
			return nfserr_stale_clientid;
		return nfs_ok;
	}

	if (STALE_CLIENTID(clid, nn))
		return nfserr_stale_clientid;

	/*
	 * For v4.1+ we get the client in the SEQUENCE op. If we don't have one
	 * cached already then we know this is for is for v4.0 and "sessions"
	 * will be false.
	 */
	WARN_ON_ONCE(cstate->session);
	found = find_confirmed_client(clid, false, nn);
	if (!found)
		return nfserr_expired;

	/* Cache the nfs4_client in cstate! */
	cstate->clp = found;
	atomic_inc(&found->cl_refcount);
	return nfs_ok;
}

3253
__be32
A
Andy Adamson 已提交
3254
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
3255
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3256 3257 3258 3259
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
3260
	struct nfs4_openowner *oo = NULL;
3261
	__be32 status;
L
Linus Torvalds 已提交
3262

3263
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
3264
		return nfserr_stale_clientid;
3265 3266 3267 3268 3269 3270 3271
	/*
	 * 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
3272

3273 3274 3275 3276 3277
	status = lookup_clientid(clientid, cstate, nn);
	if (status)
		return status;
	clp = cstate->clp;

3278
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
3279
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
3280 3281
	open->op_openowner = oo;
	if (!oo) {
3282
		goto new_owner;
L
Linus Torvalds 已提交
3283
	}
3284
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
3285
		/* Replace unconfirmed owners without checking for replay. */
3286 3287
		release_openowner(oo);
		open->op_openowner = NULL;
3288
		goto new_owner;
3289
	}
3290 3291 3292 3293
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	goto alloc_stateid;
3294
new_owner:
3295
	oo = alloc_init_open_stateowner(strhashval, open, cstate);
3296 3297 3298
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
3299
alloc_stateid:
3300
	open->op_stp = nfs4_alloc_open_stateid(clp);
3301 3302
	if (!open->op_stp)
		return nfserr_jukebox;
3303
	return nfs_ok;
L
Linus Torvalds 已提交
3304 3305
}

3306
static inline __be32
N
NeilBrown 已提交
3307 3308 3309 3310 3311 3312 3313 3314
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;
}

3315
static int share_access_to_flags(u32 share_access)
3316
{
3317
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
3318 3319
}

3320
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
3321
{
3322
	struct nfs4_stid *ret;
3323

3324
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
3325 3326 3327
	if (!ret)
		return NULL;
	return delegstateid(ret);
3328 3329
}

3330 3331 3332 3333 3334 3335
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;
}

3336
static __be32
3337
nfs4_check_deleg(struct nfs4_client *cl, struct nfsd4_open *open,
3338 3339 3340
		struct nfs4_delegation **dp)
{
	int flags;
3341
	__be32 status = nfserr_bad_stateid;
3342

3343
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
3344
	if (*dp == NULL)
3345
		goto out;
3346
	flags = share_access_to_flags(open->op_share_access);
3347 3348 3349
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
3350
out:
3351
	if (!nfsd4_is_deleg_cur(open))
3352 3353 3354
		return nfs_ok;
	if (status)
		return status;
3355
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
3356
	return nfs_ok;
3357 3358
}

3359 3360
static struct nfs4_ol_stateid *
nfsd4_find_existing_open(struct nfs4_file *fp, struct nfsd4_open *open)
L
Linus Torvalds 已提交
3361
{
3362
	struct nfs4_ol_stateid *local, *ret = NULL;
3363
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
3364

3365
	spin_lock(&fp->fi_lock);
3366
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
3367 3368 3369
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
3370
		if (local->st_stateowner == &oo->oo_owner) {
3371
			ret = local;
3372
			break;
3373
		}
L
Linus Torvalds 已提交
3374
	}
3375
	spin_unlock(&fp->fi_lock);
3376
	return ret;
L
Linus Torvalds 已提交
3377 3378
}

3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389
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;
}

3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404
static inline __be32
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))
		return nfserr_inval;
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

3405
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
3406 3407
		struct svc_fh *cur_fh, struct nfs4_ol_stateid *stp,
		struct nfsd4_open *open)
3408
{
3409
	struct file *filp = NULL;
3410
	__be32 status;
3411 3412
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);
3413
	unsigned char old_access_bmap, old_deny_bmap;
3414

3415
	spin_lock(&fp->fi_lock);
3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442

	/*
	 * Are we trying to set a deny mode that would conflict with
	 * current access?
	 */
	status = nfs4_file_check_deny(fp, open->op_share_deny);
	if (status != nfs_ok) {
		spin_unlock(&fp->fi_lock);
		goto out;
	}

	/* set access to the file */
	status = nfs4_file_get_access(fp, open->op_share_access);
	if (status != nfs_ok) {
		spin_unlock(&fp->fi_lock);
		goto out;
	}

	/* Set access bits in stateid */
	old_access_bmap = stp->st_access_bmap;
	set_access(open->op_share_access, stp);

	/* Set new deny mask */
	old_deny_bmap = stp->st_deny_bmap;
	set_deny(open->op_share_deny, stp);
	fp->fi_share_deny |= (open->op_share_deny & NFS4_SHARE_DENY_BOTH);

3443
	if (!fp->fi_fds[oflag]) {
3444 3445
		spin_unlock(&fp->fi_lock);
		status = nfsd_open(rqstp, cur_fh, S_IFREG, access, &filp);
3446
		if (status)
3447
			goto out_put_access;
3448 3449 3450 3451 3452
		spin_lock(&fp->fi_lock);
		if (!fp->fi_fds[oflag]) {
			fp->fi_fds[oflag] = filp;
			filp = NULL;
		}
3453
	}
3454 3455 3456
	spin_unlock(&fp->fi_lock);
	if (filp)
		fput(filp);
3457

3458 3459 3460 3461 3462
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status)
		goto out_put_access;
out:
	return status;
3463 3464 3465 3466 3467
out_put_access:
	stp->st_access_bmap = old_access_bmap;
	nfs4_file_put_access(fp, open->op_share_access);
	reset_union_bmap_deny(bmap_to_share_mode(old_deny_bmap), stp);
	goto out;
L
Linus Torvalds 已提交
3468 3469
}

3470
static __be32
3471
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
3472
{
3473
	__be32 status;
3474
	unsigned char old_deny_bmap;
L
Linus Torvalds 已提交
3475

3476
	if (!test_access(open->op_share_access, stp))
3477
		return nfs4_get_vfs_file(rqstp, fp, cur_fh, stp, open);
3478

3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490
	/* test and set deny mode */
	spin_lock(&fp->fi_lock);
	status = nfs4_file_check_deny(fp, open->op_share_deny);
	if (status == nfs_ok) {
		old_deny_bmap = stp->st_deny_bmap;
		set_deny(open->op_share_deny, stp);
		fp->fi_share_deny |=
				(open->op_share_deny & NFS4_SHARE_DENY_BOTH);
	}
	spin_unlock(&fp->fi_lock);

	if (status != nfs_ok)
L
Linus Torvalds 已提交
3491 3492
		return status;

3493 3494 3495 3496 3497
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status != nfs_ok)
		reset_union_bmap_deny(old_deny_bmap, stp);
	return status;
}
L
Linus Torvalds 已提交
3498 3499

static void
3500
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
3501
{
3502
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
3503 3504
}

3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517
/* 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;
}

3518
static struct file_lock *nfs4_alloc_init_lease(struct nfs4_file *fp, int flag)
3519 3520 3521 3522 3523 3524 3525 3526
{
	struct file_lock *fl;

	fl = locks_alloc_lock();
	if (!fl)
		return NULL;
	locks_init_lock(fl);
	fl->fl_lmops = &nfsd_lease_mng_ops;
3527
	fl->fl_flags = FL_DELEG;
3528 3529
	fl->fl_type = flag == NFS4_OPEN_DELEGATE_READ? F_RDLCK: F_WRLCK;
	fl->fl_end = OFFSET_MAX;
3530
	fl->fl_owner = (fl_owner_t)fp;
3531 3532 3533 3534
	fl->fl_pid = current->tgid;
	return fl;
}

3535
static int nfs4_setlease(struct nfs4_delegation *dp)
3536
{
3537
	struct nfs4_file *fp = dp->dl_stid.sc_file;
3538
	struct file_lock *fl;
3539 3540
	struct file *filp;
	int status = 0;
3541

3542
	fl = nfs4_alloc_init_lease(fp, NFS4_OPEN_DELEGATE_READ);
3543 3544
	if (!fl)
		return -ENOMEM;
3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569
	filp = find_readable_file(fp);
	if (!filp) {
		/* We should always have a readable file here */
		WARN_ON_ONCE(1);
		return -EBADF;
	}
	fl->fl_file = filp;
	status = vfs_setlease(filp, fl->fl_type, &fl);
	if (status) {
		locks_free_lock(fl);
		goto out_fput;
	}
	spin_lock(&state_lock);
	spin_lock(&fp->fi_lock);
	/* Did the lease get broken before we took the lock? */
	status = -EAGAIN;
	if (fp->fi_had_conflict)
		goto out_unlock;
	/* Race breaker */
	if (fp->fi_lease) {
		status = 0;
		atomic_inc(&fp->fi_delegees);
		hash_delegation_locked(dp, fp);
		goto out_unlock;
	}
3570
	fp->fi_lease = fl;
3571
	fp->fi_deleg_file = filp;
3572
	atomic_set(&fp->fi_delegees, 1);
3573
	hash_delegation_locked(dp, fp);
3574
	spin_unlock(&fp->fi_lock);
3575
	spin_unlock(&state_lock);
3576
	return 0;
3577 3578 3579 3580 3581
out_unlock:
	spin_unlock(&fp->fi_lock);
	spin_unlock(&state_lock);
out_fput:
	fput(filp);
3582
	return status;
3583 3584
}

J
Jeff Layton 已提交
3585 3586 3587
static struct nfs4_delegation *
nfs4_set_delegation(struct nfs4_client *clp, struct svc_fh *fh,
		    struct nfs4_file *fp)
3588
{
J
Jeff Layton 已提交
3589 3590
	int status;
	struct nfs4_delegation *dp;
3591

3592
	if (fp->fi_had_conflict)
J
Jeff Layton 已提交
3593 3594 3595 3596 3597 3598
		return ERR_PTR(-EAGAIN);

	dp = alloc_init_deleg(clp, fh);
	if (!dp)
		return ERR_PTR(-ENOMEM);

3599
	get_nfs4_file(fp);
3600 3601
	spin_lock(&state_lock);
	spin_lock(&fp->fi_lock);
3602
	dp->dl_stid.sc_file = fp;
3603 3604 3605
	if (!fp->fi_lease) {
		spin_unlock(&fp->fi_lock);
		spin_unlock(&state_lock);
J
Jeff Layton 已提交
3606 3607
		status = nfs4_setlease(dp);
		goto out;
3608
	}
3609
	atomic_inc(&fp->fi_delegees);
3610
	if (fp->fi_had_conflict) {
3611 3612
		status = -EAGAIN;
		goto out_unlock;
3613
	}
3614
	hash_delegation_locked(dp, fp);
J
Jeff Layton 已提交
3615
	status = 0;
3616 3617
out_unlock:
	spin_unlock(&fp->fi_lock);
3618
	spin_unlock(&state_lock);
J
Jeff Layton 已提交
3619 3620
out:
	if (status) {
3621
		nfs4_put_stid(&dp->dl_stid);
J
Jeff Layton 已提交
3622 3623 3624
		return ERR_PTR(status);
	}
	return dp;
3625 3626
}

3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642
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:
3643
			WARN_ON_ONCE(1);
3644 3645 3646 3647
		}
	}
}

L
Linus Torvalds 已提交
3648 3649
/*
 * Attempt to hand out a delegation.
3650 3651 3652
 *
 * Note we don't support write delegations, and won't until the vfs has
 * proper support for them.
L
Linus Torvalds 已提交
3653 3654
 */
static void
3655 3656
nfs4_open_delegation(struct svc_fh *fh, struct nfsd4_open *open,
			struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3657 3658
{
	struct nfs4_delegation *dp;
3659 3660
	struct nfs4_openowner *oo = openowner(stp->st_stateowner);
	struct nfs4_client *clp = stp->st_stid.sc_client;
3661
	int cb_up;
3662
	int status = 0;
L
Linus Torvalds 已提交
3663

3664
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
3665 3666 3667
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
3668
			if (!cb_up)
3669
				open->op_recall = 1;
3670 3671
			if (open->op_delegate_type != NFS4_OPEN_DELEGATE_READ)
				goto out_no_deleg;
3672 3673
			break;
		case NFS4_OPEN_CLAIM_NULL:
3674
		case NFS4_OPEN_CLAIM_FH:
3675 3676 3677 3678
			/*
			 * Let's not give out any delegations till everyone's
			 * had the chance to reclaim theirs....
			 */
3679
			if (locks_in_grace(clp->net))
3680
				goto out_no_deleg;
3681
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
3682
				goto out_no_deleg;
3683 3684 3685 3686 3687 3688 3689
			/*
			 * Also, if the file was opened for write or
			 * create, there's a good chance the client's
			 * about to write to it, resulting in an
			 * immediate recall (since we don't support
			 * write delegations):
			 */
3690
			if (open->op_share_access & NFS4_SHARE_ACCESS_WRITE)
3691 3692 3693
				goto out_no_deleg;
			if (open->op_create == NFS4_OPEN_CREATE)
				goto out_no_deleg;
3694 3695
			break;
		default:
3696
			goto out_no_deleg;
3697
	}
3698
	dp = nfs4_set_delegation(clp, fh, stp->st_stid.sc_file);
J
Jeff Layton 已提交
3699
	if (IS_ERR(dp))
3700
		goto out_no_deleg;
L
Linus Torvalds 已提交
3701

3702
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
3703

3704
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
3705
		STATEID_VAL(&dp->dl_stid.sc_stateid));
3706
	open->op_delegate_type = NFS4_OPEN_DELEGATE_READ;
3707 3708
	return;
out_no_deleg:
3709 3710
	open->op_delegate_type = NFS4_OPEN_DELEGATE_NONE;
	if (open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS &&
3711
	    open->op_delegate_type != NFS4_OPEN_DELEGATE_NONE) {
3712
		dprintk("NFSD: WARNING: refusing delegation reclaim\n");
3713 3714
		open->op_recall = 1;
	}
3715 3716 3717 3718 3719

	/* 4.1 client asking for a delegation? */
	if (open->op_deleg_want)
		nfsd4_open_deleg_none_ext(open, status);
	return;
L
Linus Torvalds 已提交
3720 3721
}

3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739
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
3740 3741 3742
/*
 * called with nfs4_lock_state() held.
 */
3743
__be32
L
Linus Torvalds 已提交
3744 3745
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3746
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3747
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3748
	struct nfs4_file *fp = NULL;
3749
	struct nfs4_ol_stateid *stp = NULL;
3750
	struct nfs4_delegation *dp = NULL;
3751
	__be32 status;
L
Linus Torvalds 已提交
3752 3753 3754 3755 3756 3757

	/*
	 * 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
	 */
3758
	fp = find_or_add_file(open->op_file, &current_fh->fh_handle);
3759
	if (fp != open->op_file) {
3760
		status = nfs4_check_deleg(cl, open, &dp);
3761 3762
		if (status)
			goto out;
3763
		stp = nfsd4_find_existing_open(fp, open);
L
Linus Torvalds 已提交
3764
	} else {
3765
		open->op_file = NULL;
3766
		status = nfserr_bad_stateid;
3767
		if (nfsd4_is_deleg_cur(open))
3768
			goto out;
3769
		status = nfserr_jukebox;
L
Linus Torvalds 已提交
3770 3771 3772 3773 3774 3775 3776 3777
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3778
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3779 3780 3781
		if (status)
			goto out;
	} else {
3782 3783
		stp = open->op_stp;
		open->op_stp = NULL;
3784
		init_open_stateid(stp, fp, open);
3785 3786 3787 3788 3789
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, stp, open);
		if (status) {
			release_open_stateid(stp);
			goto out;
		}
L
Linus Torvalds 已提交
3790
	}
3791 3792
	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
3793

3794 3795 3796 3797 3798 3799 3800 3801
	if (nfsd4_has_session(&resp->cstate)) {
		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
3802 3803 3804 3805
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3806
	nfs4_open_delegation(current_fh, open, stp);
3807
nodeleg:
L
Linus Torvalds 已提交
3808 3809
	status = nfs_ok;

3810
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3811
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3812
out:
3813 3814
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3815 3816
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3817

3818 3819
	if (fp)
		put_nfs4_file(fp);
3820
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3821
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3822 3823 3824 3825
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3826
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3827
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3828 3829 3830 3831 3832
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847
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;
		}
	}
3848 3849
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3850
	if (open->op_stp)
3851
		nfs4_put_stid(&open->op_stp->st_stid);
3852 3853
}

3854
__be32
3855 3856
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3857 3858
{
	struct nfs4_client *clp;
3859
	__be32 status;
3860
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3861 3862 3863 3864

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3865
	status = lookup_clientid(clid, cstate, nn);
3866
	if (status)
L
Linus Torvalds 已提交
3867
		goto out;
3868
	clp = cstate->clp;
L
Linus Torvalds 已提交
3869
	status = nfserr_cb_path_down;
3870
	if (!list_empty(&clp->cl_delegations)
3871
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3872 3873 3874 3875 3876 3877 3878
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3879
static void
3880
nfsd4_end_grace(struct nfsd_net *nn)
3881
{
3882
	/* do nothing if grace period already ended */
3883
	if (nn->grace_ended)
3884 3885
		return;

3886
	dprintk("NFSD: end of grace period\n");
3887
	nn->grace_ended = true;
3888
	nfsd4_record_grace_done(nn, nn->boot_time);
3889
	locks_end_grace(&nn->nfsd4_manager);
3890 3891 3892 3893 3894
	/*
	 * 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:
	 */
3895
	nn->nfsd4_grace = nn->nfsd4_lease;
3896 3897
}

3898
static time_t
3899
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3900 3901
{
	struct nfs4_client *clp;
3902
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3903 3904
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3905
	time_t cutoff = get_seconds() - nn->nfsd4_lease;
3906
	time_t t, new_timeo = nn->nfsd4_lease;
L
Linus Torvalds 已提交
3907 3908 3909 3910

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3911
	nfsd4_end_grace(nn);
3912
	INIT_LIST_HEAD(&reaplist);
3913
	spin_lock(&nn->client_lock);
3914
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3915 3916 3917
		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;
3918
			new_timeo = min(new_timeo, t);
L
Linus Torvalds 已提交
3919 3920
			break;
		}
3921
		if (mark_client_expired_locked(clp)) {
3922 3923 3924 3925
			dprintk("NFSD: client in use (clientid %08x)\n",
				clp->cl_clientid.cl_id);
			continue;
		}
3926
		list_move(&clp->cl_lru, &reaplist);
3927
	}
3928
	spin_unlock(&nn->client_lock);
3929 3930
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3931 3932 3933 3934
		dprintk("NFSD: purging unused client (clientid %08x)\n",
			clp->cl_clientid.cl_id);
		expire_client(clp);
	}
3935
	spin_lock(&state_lock);
3936
	list_for_each_safe(pos, next, &nn->del_recall_lru) {
L
Linus Torvalds 已提交
3937
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
3938 3939
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3940
		if (time_after((unsigned long)dp->dl_time, (unsigned long)cutoff)) {
3941 3942
			t = dp->dl_time - cutoff;
			new_timeo = min(new_timeo, t);
L
Linus Torvalds 已提交
3943 3944
			break;
		}
3945 3946
		unhash_delegation_locked(dp);
		list_add(&dp->dl_recall_lru, &reaplist);
L
Linus Torvalds 已提交
3947
	}
3948
	spin_unlock(&state_lock);
3949 3950 3951 3952
	while (!list_empty(&reaplist)) {
		dp = list_first_entry(&reaplist, struct nfs4_delegation,
					dl_recall_lru);
		list_del_init(&dp->dl_recall_lru);
3953
		revoke_delegation(dp);
L
Linus Torvalds 已提交
3954
	}
3955
	list_for_each_safe(pos, next, &nn->close_lru) {
3956 3957
		oo = container_of(pos, struct nfs4_openowner, oo_close_lru);
		if (time_after((unsigned long)oo->oo_time, (unsigned long)cutoff)) {
3958 3959
			t = oo->oo_time - cutoff;
			new_timeo = min(new_timeo, t);
L
Linus Torvalds 已提交
3960 3961
			break;
		}
3962
		release_openowner(oo);
L
Linus Torvalds 已提交
3963
	}
3964
	new_timeo = max_t(time_t, new_timeo, NFSD_LAUNDROMAT_MINTIMEOUT);
L
Linus Torvalds 已提交
3965
	nfs4_unlock_state();
3966
	return new_timeo;
L
Linus Torvalds 已提交
3967 3968
}

H
Harvey Harrison 已提交
3969 3970 3971 3972
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3973
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3974 3975
{
	time_t t;
3976 3977 3978 3979
	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
3980

3981
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3982
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3983
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3984 3985
}

3986
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3987
{
3988
	if (!nfsd_fh_match(&fhp->fh_handle, &stp->st_stid.sc_file->fi_fhandle))
3989 3990
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3991 3992 3993
}

static inline int
3994
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3995
{
3996 3997 3998
	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
3999 4000 4001
}

static inline int
4002
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
4003
{
4004 4005
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
4006 4007 4008
}

static
4009
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
4010
{
4011
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
4012

4013 4014 4015
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
4016
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
4017
                goto out;
4018
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
4019 4020 4021 4022 4023 4024
                goto out;
	status = nfs_ok;
out:
	return status;
}

4025
static inline __be32
4026
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
4027
{
4028
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
4029
		return nfs_ok;
4030
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
4031
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
		 * 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
4047
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
4048
{
4049
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
4050 4051
}

4052 4053 4054
/* Returns true iff a is later than b: */
static bool stateid_generation_after(stateid_t *a, stateid_t *b)
{
J
Jim Rees 已提交
4055
	return (s32)(a->si_generation - b->si_generation) > 0;
4056 4057
}

J
J. Bruce Fields 已提交
4058
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
4059
{
A
Andy Adamson 已提交
4060 4061 4062 4063
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
4064
	if (has_session && in->si_generation == 0)
4065 4066 4067 4068
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
4069

4070
	/* If the client sends us a stateid from the future, it's buggy: */
4071
	if (stateid_generation_after(in, ref))
4072 4073
		return nfserr_bad_stateid;
	/*
4074 4075 4076 4077 4078 4079 4080 4081
	 * 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:
4082
	 */
4083
	return nfserr_old_stateid;
4084 4085
}

4086
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
4087
{
4088 4089
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
4090
	__be32 status = nfserr_bad_stateid;
4091

4092
	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
4093
		return status;
4094 4095 4096 4097 4098 4099 4100
	/* 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);
4101
		return status;
4102
	}
4103 4104
	spin_lock(&cl->cl_lock);
	s = find_stateid_locked(cl, stateid);
4105
	if (!s)
4106
		goto out_unlock;
4107
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
4108
	if (status)
4109
		goto out_unlock;
4110 4111
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
4112 4113
		status = nfs_ok;
		break;
4114
	case NFS4_REVOKED_DELEG_STID:
4115 4116
		status = nfserr_deleg_revoked;
		break;
4117 4118 4119 4120 4121 4122
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
		ols = openlockstateid(s);
		if (ols->st_stateowner->so_is_open_owner
	    			&& !(openowner(ols->st_stateowner)->oo_flags
						& NFS4_OO_CONFIRMED))
4123 4124 4125 4126
			status = nfserr_bad_stateid;
		else
			status = nfs_ok;
		break;
4127 4128
	default:
		printk("unknown stateid type %x\n", s->sc_type);
4129
		/* Fallthrough */
4130
	case NFS4_CLOSED_STID:
4131
	case NFS4_CLOSED_DELEG_STID:
4132
		status = nfserr_bad_stateid;
4133
	}
4134 4135 4136
out_unlock:
	spin_unlock(&cl->cl_lock);
	return status;
4137 4138
}

4139 4140 4141 4142
static __be32
nfsd4_lookup_stateid(struct nfsd4_compound_state *cstate,
		     stateid_t *stateid, unsigned char typemask,
		     struct nfs4_stid **s, struct nfsd_net *nn)
4143
{
J
J. Bruce Fields 已提交
4144
	__be32 status;
4145 4146 4147

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
4148
	status = lookup_clientid(&stateid->si_opaque.so_clid, cstate, nn);
4149
	if (status == nfserr_stale_clientid) {
4150
		if (cstate->session)
4151
			return nfserr_bad_stateid;
4152
		return nfserr_stale_stateid;
4153
	}
J
J. Bruce Fields 已提交
4154 4155
	if (status)
		return status;
4156
	*s = find_stateid_by_type(cstate->clp, stateid, typemask);
4157 4158 4159 4160 4161
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;
}

L
Linus Torvalds 已提交
4162 4163 4164
/*
* Checks for stateid operations
*/
4165
__be32
4166
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
4167
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
4168
{
4169
	struct nfs4_stid *s;
4170
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
4171
	struct nfs4_delegation *dp = NULL;
4172
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
4173
	struct inode *ino = current_fh->fh_dentry->d_inode;
4174
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4175
	struct file *file = NULL;
4176
	__be32 status;
L
Linus Torvalds 已提交
4177 4178 4179 4180

	if (filpp)
		*filpp = NULL;

4181
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
4182 4183 4184
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
4185
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
4186

4187 4188
	nfs4_lock_state();

4189
	status = nfsd4_lookup_stateid(cstate, stateid,
4190
				NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
4191
				&s, nn);
4192
	if (status)
4193
		goto out;
4194 4195 4196
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
4197 4198
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
4199
		dp = delegstateid(s);
4200 4201 4202
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
4203
		if (filpp) {
4204
			file = dp->dl_stid.sc_file->fi_deleg_file;
4205
			if (!file) {
4206 4207 4208 4209
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
4210
			get_file(file);
4211
		}
4212 4213 4214
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
4215
		stp = openlockstateid(s);
4216 4217
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
4218
			goto out;
4219
		if (stp->st_stateowner->so_is_open_owner
4220
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
4221
			goto out;
4222 4223
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
4224
			goto out;
4225
		if (filpp) {
4226 4227
			struct nfs4_file *fp = stp->st_stid.sc_file;

4228
			if (flags & RD_STATE)
4229
				file = find_readable_file(fp);
4230
			else
4231
				file = find_writeable_file(fp);
4232
		}
4233 4234
		break;
	default:
4235 4236
		status = nfserr_bad_stateid;
		goto out;
L
Linus Torvalds 已提交
4237 4238
	}
	status = nfs_ok;
4239
	if (file)
4240
		*filpp = file;
L
Linus Torvalds 已提交
4241
out:
4242
	nfs4_unlock_state();
L
Linus Torvalds 已提交
4243 4244 4245
	return status;
}

4246
static __be32
4247
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
4248
{
4249 4250
	struct nfs4_lockowner *lo = lockowner(stp->st_stateowner);

4251
	if (check_for_locks(stp->st_stid.sc_file, lo))
4252
		return nfserr_locks_held;
4253
	release_lockowner_if_empty(lo);
4254 4255 4256
	return nfs_ok;
}

4257 4258 4259 4260 4261 4262 4263
/*
 * 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)
{
4264 4265 4266 4267 4268
	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)
4269 4270
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
4271 4272
	nfs4_unlock_state();

4273 4274 4275
	return nfs_ok;
}

4276 4277 4278 4279 4280
__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
4281
	struct nfs4_stid *s;
4282
	struct nfs4_delegation *dp;
4283
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
4284
	__be32 ret = nfserr_bad_stateid;
4285 4286

	nfs4_lock_state();
4287 4288
	spin_lock(&cl->cl_lock);
	s = find_stateid_locked(cl, stateid);
J
J. Bruce Fields 已提交
4289
	if (!s)
4290
		goto out_unlock;
J
J. Bruce Fields 已提交
4291 4292
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
4293
		ret = nfserr_locks_held;
4294
		break;
J
J. Bruce Fields 已提交
4295 4296 4297
	case NFS4_OPEN_STID:
		ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
		if (ret)
4298 4299
			break;
		ret = nfserr_locks_held;
4300
		break;
4301 4302 4303 4304 4305 4306 4307
	case NFS4_LOCK_STID:
		ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
		if (ret)
			break;
		spin_unlock(&cl->cl_lock);
		ret = nfsd4_free_lock_stateid(openlockstateid(s));
		goto out;
4308 4309
	case NFS4_REVOKED_DELEG_STID:
		dp = delegstateid(s);
4310 4311
		list_del_init(&dp->dl_recall_lru);
		spin_unlock(&cl->cl_lock);
4312
		nfs4_put_stid(s);
4313
		ret = nfs_ok;
4314 4315
		goto out;
	/* Default falls through and returns nfserr_bad_stateid */
4316
	}
4317 4318
out_unlock:
	spin_unlock(&cl->cl_lock);
4319 4320 4321 4322 4323
out:
	nfs4_unlock_state();
	return ret;
}

4324 4325 4326 4327 4328 4329
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
4330

4331
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
4332 4333 4334 4335 4336 4337 4338 4339
{
	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;
4340 4341
	if (stp->st_stid.sc_type == NFS4_CLOSED_STID
		|| stp->st_stid.sc_type == NFS4_REVOKED_DELEG_STID)
4342 4343
		/*
		 * "Closed" stateid's exist *only* to return
4344 4345
		 * nfserr_replay_me from the previous step, and
		 * revoked delegations are kept only for free_stateid.
4346 4347 4348 4349 4350 4351
		 */
		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);
4352 4353
}

L
Linus Torvalds 已提交
4354 4355 4356
/* 
 * Checks for sequence id mutating operations. 
 */
4357
static __be32
4358
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
4359
			 stateid_t *stateid, char typemask,
4360 4361
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
4362
{
4363
	__be32 status;
4364
	struct nfs4_stid *s;
4365
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
4366

4367 4368
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
4369

L
Linus Torvalds 已提交
4370
	*stpp = NULL;
4371
	status = nfsd4_lookup_stateid(cstate, stateid, typemask, &s, nn);
4372 4373
	if (status)
		return status;
4374
	stp = openlockstateid(s);
4375
	if (!nfsd4_has_session(cstate))
4376
		cstate->replay_owner = stp->st_stateowner;
L
Linus Torvalds 已提交
4377

4378 4379 4380 4381
	status = nfs4_seqid_op_checks(cstate, stateid, seqid, stp);
	if (!status)
		*stpp = stp;
	return status;
4382
}
4383

4384 4385
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)
4386 4387 4388
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
4389

4390
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
4391
						NFS4_OPEN_STID, stpp, nn);
4392 4393
	if (status)
		return status;
4394
	oo = openowner((*stpp)->st_stateowner);
4395
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
4396 4397
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
4398 4399
}

4400
__be32
4401
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4402
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
4403
{
4404
	__be32 status;
4405
	struct nfs4_openowner *oo;
4406
	struct nfs4_ol_stateid *stp;
4407
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4408

A
Al Viro 已提交
4409 4410
	dprintk("NFSD: nfsd4_open_confirm on file %pd\n",
			cstate->current_fh.fh_dentry);
L
Linus Torvalds 已提交
4411

4412
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
4413 4414
	if (status)
		return status;
L
Linus Torvalds 已提交
4415 4416 4417

	nfs4_lock_state();

4418
	status = nfs4_preprocess_seqid_op(cstate,
4419
					oc->oc_seqid, &oc->oc_req_stateid,
4420
					NFS4_OPEN_STID, &stp, nn);
4421
	if (status)
4422
		goto out;
4423
	oo = openowner(stp->st_stateowner);
4424
	status = nfserr_bad_stateid;
4425
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
4426
		goto out;
4427
	oo->oo_flags |= NFS4_OO_CONFIRMED;
4428 4429
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
4430
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
4431
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
4432

4433
	nfsd4_client_record_create(oo->oo_owner.so_client);
4434
	status = nfs_ok;
L
Linus Torvalds 已提交
4435
out:
4436
	nfsd4_bump_seqid(cstate, status);
4437 4438
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
4439 4440 4441
	return status;
}

J
J. Bruce Fields 已提交
4442
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
4443
{
4444
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
4445
		return;
4446
	nfs4_file_put_access(stp->st_stid.sc_file, access);
4447
	clear_access(access, stp);
J
J. Bruce Fields 已提交
4448
}
4449

J
J. Bruce Fields 已提交
4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
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:
4464
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
4465 4466 4467
	}
}

4468
__be32
4469 4470
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
4471
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
4472
{
4473
	__be32 status;
4474
	struct nfs4_ol_stateid *stp;
4475
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4476

A
Al Viro 已提交
4477 4478
	dprintk("NFSD: nfsd4_open_downgrade on file %pd\n", 
			cstate->current_fh.fh_dentry);
L
Linus Torvalds 已提交
4479

4480
	/* We don't yet support WANT bits: */
4481 4482 4483
	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
4484 4485

	nfs4_lock_state();
4486
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
4487
					&od->od_stateid, &stp, nn);
4488
	if (status)
L
Linus Torvalds 已提交
4489 4490
		goto out; 
	status = nfserr_inval;
4491
	if (!test_access(od->od_share_access, stp)) {
4492
		dprintk("NFSD: access not a subset of current bitmap: 0x%hhx, input access=%08x\n",
L
Linus Torvalds 已提交
4493 4494 4495
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
4496
	if (!test_deny(od->od_share_deny, stp)) {
4497
		dprintk("NFSD: deny not a subset of current bitmap: 0x%hhx, input deny=%08x\n",
L
Linus Torvalds 已提交
4498 4499 4500
			stp->st_deny_bmap, od->od_share_deny);
		goto out;
	}
J
J. Bruce Fields 已提交
4501
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
4502

4503
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
4504

4505 4506
	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
4507 4508
	status = nfs_ok;
out:
4509
	nfsd4_bump_seqid(cstate, status);
4510 4511
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
4512 4513 4514
	return status;
}

4515 4516
static void nfsd4_close_open_stateid(struct nfs4_ol_stateid *s)
{
4517 4518 4519
	struct nfs4_client *clp = s->st_stid.sc_client;
	struct nfs4_openowner *oo = openowner(s->st_stateowner);

4520
	s->st_stid.sc_type = NFS4_CLOSED_STID;
4521 4522 4523 4524 4525
	unhash_open_stateid(s);

	if (clp->cl_minorversion) {
		if (list_empty(&oo->oo_owner.so_stateids))
			release_openowner(oo);
4526
		nfs4_put_stid(&s->st_stid);
4527
	} else {
4528 4529 4530 4531 4532 4533 4534
		/*
		 * In the 4.0 case we need to keep the owners around a
		 * little while to handle CLOSE replay. We still do need
		 * to release any file access that is held by them
		 * before returning however.
		 */
		release_all_access(s);
4535 4536 4537
		if (s->st_stid.sc_file) {
			put_nfs4_file(s->st_stid.sc_file);
			s->st_stid.sc_file = NULL;
4538
		}
4539 4540 4541 4542
		oo->oo_last_closed_stid = s;
		if (list_empty(&oo->oo_owner.so_stateids))
			move_to_close_lru(oo, clp->net);
	}
4543 4544
}

L
Linus Torvalds 已提交
4545 4546 4547
/*
 * nfs4_unlock_state() called after encode
 */
4548
__be32
4549
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4550
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
4551
{
4552
	__be32 status;
4553
	struct nfs4_ol_stateid *stp;
4554 4555
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4556

A
Al Viro 已提交
4557 4558
	dprintk("NFSD: nfsd4_close on file %pd\n", 
			cstate->current_fh.fh_dentry);
L
Linus Torvalds 已提交
4559 4560

	nfs4_lock_state();
4561 4562 4563
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
4564
					&stp, nn);
4565
	nfsd4_bump_seqid(cstate, status);
4566
	if (status)
L
Linus Torvalds 已提交
4567
		goto out; 
4568 4569
	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
4570

4571
	nfsd4_close_open_stateid(stp);
L
Linus Torvalds 已提交
4572
out:
4573 4574
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
4575 4576 4577
	return status;
}

4578
__be32
4579 4580
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
4581
{
4582 4583
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
4584
	struct nfs4_stid *s;
4585
	__be32 status;
4586
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4587

4588
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
4589
		return status;
L
Linus Torvalds 已提交
4590 4591

	nfs4_lock_state();
4592
	status = nfsd4_lookup_stateid(cstate, stateid, NFS4_DELEG_STID, &s, nn);
4593
	if (status)
4594
		goto out;
4595
	dp = delegstateid(s);
4596
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
4597 4598 4599
	if (status)
		goto out;

4600
	destroy_delegation(dp);
L
Linus Torvalds 已提交
4601
out:
4602 4603
	nfs4_unlock_state();

L
Linus Torvalds 已提交
4604 4605 4606 4607 4608 4609
	return status;
}


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

B
Benny Halevy 已提交
4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624
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;

4625
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
4626 4627 4628 4629
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

L
Linus Torvalds 已提交
4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646
/*
 * 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;
}

4647 4648
/* 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. */
4649
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
4650
};
L
Linus Torvalds 已提交
4651 4652 4653 4654

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
4655
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4656

4657
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
4658 4659 4660
		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);
4661 4662 4663
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
4664 4665
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
4666
	} else {
4667 4668 4669
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
4670 4671
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
4672 4673
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
4674 4675
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
4676 4677 4678 4679 4680 4681
		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;
}

4682
static struct nfs4_lockowner *
4683 4684
find_lockowner_str(clientid_t *clid, struct xdr_netobj *owner,
		struct nfsd_net *nn)
L
Linus Torvalds 已提交
4685
{
4686 4687
	unsigned int strhashval = ownerstr_hashval(clid->cl_id, owner);
	struct nfs4_stateowner *so;
L
Linus Torvalds 已提交
4688

4689 4690 4691 4692 4693 4694
	list_for_each_entry(so, &nn->ownerstr_hashtbl[strhashval], so_strhash) {
		if (so->so_is_open_owner)
			continue;
		if (!same_owner_str(so, owner, clid))
			continue;
		return lockowner(so);
L
Linus Torvalds 已提交
4695 4696 4697 4698 4699 4700 4701
	}
	return NULL;
}

/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4702
 * occurred. 
L
Linus Torvalds 已提交
4703
 *
4704
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4705
 */
4706
static struct nfs4_lockowner *
4707
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4708
	struct nfs4_lockowner *lo;
4709
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
4710

4711 4712
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4713
		return NULL;
4714 4715
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4716 4717
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4718
	lo->lo_owner.so_seqid = lock->lk_new_lock_seqid + 1;
4719
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
4720
	return lo;
L
Linus Torvalds 已提交
4721 4722
}

4723 4724 4725 4726
static void
init_lock_stateid(struct nfs4_ol_stateid *stp, struct nfs4_lockowner *lo,
		  struct nfs4_file *fp, struct inode *inode,
		  struct nfs4_ol_stateid *open_stp)
L
Linus Torvalds 已提交
4727
{
4728
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4729

4730 4731
	lockdep_assert_held(&clp->cl_lock);

4732
	atomic_inc(&stp->st_stid.sc_count);
J
J. Bruce Fields 已提交
4733
	stp->st_stid.sc_type = NFS4_LOCK_STID;
4734
	stp->st_stateowner = &lo->lo_owner;
4735
	get_nfs4_file(fp);
4736
	stp->st_stid.sc_file = fp;
4737
	stp->st_stid.sc_free = nfs4_free_lock_stateid;
J
J. Bruce Fields 已提交
4738
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4739
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4740
	stp->st_openstp = open_stp;
4741
	list_add(&stp->st_locks, &open_stp->st_locks);
4742
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
4743 4744 4745
	spin_lock(&fp->fi_lock);
	list_add(&stp->st_perfile, &fp->fi_stateids);
	spin_unlock(&fp->fi_lock);
L
Linus Torvalds 已提交
4746 4747
}

4748 4749 4750 4751
static struct nfs4_ol_stateid *
find_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp)
{
	struct nfs4_ol_stateid *lst;
4752 4753 4754
	struct nfs4_client *clp = lo->lo_owner.so_client;

	lockdep_assert_held(&clp->cl_lock);
4755 4756

	list_for_each_entry(lst, &lo->lo_owner.so_stateids, st_perstateowner) {
4757 4758
		if (lst->st_stid.sc_file == fp) {
			atomic_inc(&lst->st_stid.sc_count);
4759
			return lst;
4760
		}
4761 4762 4763 4764
	}
	return NULL;
}

4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796
static struct nfs4_ol_stateid *
find_or_create_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fi,
			    struct inode *inode, struct nfs4_ol_stateid *ost,
			    bool *new)
{
	struct nfs4_stid *ns = NULL;
	struct nfs4_ol_stateid *lst;
	struct nfs4_openowner *oo = openowner(ost->st_stateowner);
	struct nfs4_client *clp = oo->oo_owner.so_client;

	spin_lock(&clp->cl_lock);
	lst = find_lock_stateid(lo, fi);
	if (lst == NULL) {
		spin_unlock(&clp->cl_lock);
		ns = nfs4_alloc_stid(clp, stateid_slab);
		if (ns == NULL)
			return NULL;

		spin_lock(&clp->cl_lock);
		lst = find_lock_stateid(lo, fi);
		if (likely(!lst)) {
			lst = openlockstateid(ns);
			init_lock_stateid(lst, lo, fi, inode, ost);
			ns = NULL;
			*new = true;
		}
	}
	spin_unlock(&clp->cl_lock);
	if (ns)
		nfs4_put_stid(ns);
	return lst;
}
4797

4798
static int
L
Linus Torvalds 已提交
4799 4800
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4801
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4802 4803 4804
	     LOFF_OVERFLOW(offset, length)));
}

4805
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4806
{
4807
	struct nfs4_file *fp = lock_stp->st_stid.sc_file;
J
J. Bruce Fields 已提交
4808

4809 4810
	lockdep_assert_held(&fp->fi_lock);

4811
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4812
		return;
4813
	__nfs4_file_get_access(fp, access);
4814
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4815 4816
}

4817 4818 4819 4820 4821
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)
4822
{
4823
	struct nfs4_file *fi = ost->st_stid.sc_file;
4824 4825
	struct nfs4_openowner *oo = openowner(ost->st_stateowner);
	struct nfs4_client *cl = oo->oo_owner.so_client;
4826
	struct inode *inode = cstate->current_fh.fh_dentry->d_inode;
4827 4828
	struct nfs4_lockowner *lo;
	unsigned int strhashval;
4829
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4830

4831
	lo = find_lockowner_str(&cl->cl_clientid, &lock->v.new.owner, nn);
4832 4833 4834 4835 4836 4837 4838 4839 4840 4841
	if (!lo) {
		strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
				&lock->v.new.owner);
		lo = alloc_init_lock_stateowner(strhashval, cl, ost, lock);
		if (lo == NULL)
			return nfserr_jukebox;
	} else {
		/* with an existing lockowner, seqids must be the same */
		if (!cstate->minorversion &&
		    lock->lk_new_lock_seqid != lo->lo_owner.so_seqid)
4842 4843
			return nfserr_bad_seqid;
	}
4844

4845
	*lst = find_or_create_lock_stateid(lo, fi, inode, ost, new);
4846
	if (*lst == NULL) {
4847 4848
		release_lockowner_if_empty(lo);
		return nfserr_jukebox;
4849 4850 4851 4852
	}
	return nfs_ok;
}

L
Linus Torvalds 已提交
4853 4854 4855
/*
 *  LOCK operation 
 */
4856
__be32
4857
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4858
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4859
{
4860 4861
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4862
	struct nfs4_ol_stateid *lock_stp = NULL;
4863
	struct nfs4_file *fp;
4864
	struct file *filp = NULL;
4865 4866
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4867
	__be32 status = 0;
4868
	bool new_state = false;
4869
	int lkflg;
4870
	int err;
4871 4872
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4873 4874 4875 4876 4877 4878 4879 4880

	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;

4881
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4882
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4883 4884 4885 4886
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4887 4888 4889
	nfs4_lock_state();

	if (lock->lk_is_new) {
4890
		struct nfs4_ol_stateid *open_stp = NULL;
4891 4892 4893 4894 4895 4896 4897

		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
4898
		status = nfserr_stale_clientid;
4899
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4900 4901 4902
			goto out;

		/* validate and update open stateid and open seqid */
4903
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4904 4905
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4906
					&open_stp, nn);
4907
		if (status)
L
Linus Torvalds 已提交
4908
			goto out;
4909
		open_sop = openowner(open_stp->st_stateowner);
4910
		status = nfserr_bad_stateid;
4911
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4912 4913
						&lock->v.new.clientid))
			goto out;
4914 4915
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
4916
	} else {
4917
		status = nfs4_preprocess_seqid_op(cstate,
4918 4919
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4920
				       NFS4_LOCK_STID, &lock_stp, nn);
4921 4922 4923 4924
		/* FIXME: move into nfs4_preprocess_seqid_op */
		if (!status)
			atomic_inc(&lock_stp->st_stid.sc_count);
	}
J
J. Bruce Fields 已提交
4925 4926
	if (status)
		goto out;
4927
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4928

4929 4930 4931 4932 4933
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4934
	status = nfserr_grace;
4935
	if (locks_in_grace(net) && !lock->lk_reclaim)
4936 4937
		goto out;
	status = nfserr_no_grace;
4938
	if (!locks_in_grace(net) && lock->lk_reclaim)
4939 4940
		goto out;

4941 4942 4943 4944 4945 4946 4947
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}

4948
	fp = lock_stp->st_stid.sc_file;
4949
	locks_init_lock(file_lock);
L
Linus Torvalds 已提交
4950 4951 4952
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4953 4954
			spin_lock(&fp->fi_lock);
			filp = find_readable_file_locked(fp);
J
J. Bruce Fields 已提交
4955 4956
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4957
			spin_unlock(&fp->fi_lock);
4958
			file_lock->fl_type = F_RDLCK;
4959
			break;
L
Linus Torvalds 已提交
4960 4961
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4962 4963
			spin_lock(&fp->fi_lock);
			filp = find_writeable_file_locked(fp);
J
J. Bruce Fields 已提交
4964 4965
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4966
			spin_unlock(&fp->fi_lock);
4967
			file_lock->fl_type = F_WRLCK;
4968
			break;
L
Linus Torvalds 已提交
4969 4970 4971 4972
		default:
			status = nfserr_inval;
		goto out;
	}
4973 4974 4975 4976
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991
	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
4992

4993
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4994
	switch (-err) {
L
Linus Torvalds 已提交
4995
	case 0: /* success! */
4996 4997
		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
4998
				sizeof(stateid_t));
4999
		status = 0;
5000 5001 5002 5003
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
5004
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
5005
		break;
L
Linus Torvalds 已提交
5006 5007
	case (EDEADLK):
		status = nfserr_deadlock;
5008
		break;
5009
	default:
5010
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
5011
		status = nfserrno(err);
5012
		break;
L
Linus Torvalds 已提交
5013 5014
	}
out:
5015 5016
	if (filp)
		fput(filp);
5017 5018
	if (lock_stp)
		nfs4_put_stid(&lock_stp->st_stid);
5019
	if (status && new_state)
5020
		release_lock_stateid(lock_stp);
5021
	nfsd4_bump_seqid(cstate, status);
5022 5023
	if (!cstate->replay_owner)
		nfs4_unlock_state();
5024 5025 5026 5027
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
5028 5029 5030
	return status;
}

5031 5032 5033 5034 5035 5036
/*
 * 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.)
 */
5037
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
5038 5039
{
	struct file *file;
5040 5041 5042 5043 5044
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
5045 5046 5047
	return err;
}

L
Linus Torvalds 已提交
5048 5049 5050
/*
 * LOCKT operation
 */
5051
__be32
5052 5053
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
5054
{
5055
	struct file_lock *file_lock = NULL;
5056
	struct nfs4_lockowner *lo;
5057
	__be32 status;
5058
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
5059

5060
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
5061 5062 5063 5064 5065 5066 5067
		return nfserr_grace;

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

	nfs4_lock_state();

5068
	if (!nfsd4_has_session(cstate)) {
5069
		status = lookup_clientid(&lockt->lt_clientid, cstate, nn);
5070 5071 5072
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
5073

5074
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
5075 5076
		goto out;

5077 5078 5079 5080 5081 5082 5083
	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
5084 5085 5086
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
5087
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
5088 5089 5090
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
5091
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
5092 5093
		break;
		default:
5094
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
5095 5096 5097 5098
			status = nfserr_inval;
		goto out;
	}

5099
	lo = find_lockowner_str(&lockt->lt_clientid, &lockt->lt_owner, nn);
5100
	if (lo)
5101 5102 5103
		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
5104

5105 5106
	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
5107

5108
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
5109

5110
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
5111
	if (status)
5112
		goto out;
5113

5114
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
5115
		status = nfserr_denied;
5116
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
5117 5118 5119
	}
out:
	nfs4_unlock_state();
5120 5121
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
5122 5123 5124
	return status;
}

5125
__be32
5126
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
5127
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
5128
{
5129
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
5130
	struct file *filp = NULL;
5131
	struct file_lock *file_lock = NULL;
5132
	__be32 status;
5133
	int err;
5134 5135
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
5136 5137 5138 5139 5140 5141 5142 5143 5144
	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();
									        
5145
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
5146 5147
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
5148
	if (status)
L
Linus Torvalds 已提交
5149
		goto out;
5150
	filp = find_any_file(stp->st_stid.sc_file);
5151 5152 5153 5154
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
5155 5156 5157 5158
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
5159
		goto fput;
5160 5161 5162
	}
	locks_init_lock(file_lock);
	file_lock->fl_type = F_UNLCK;
5163
	file_lock->fl_owner = (fl_owner_t)lockowner(stp->st_stateowner);
5164 5165 5166 5167 5168 5169 5170 5171 5172
	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
5173

5174
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
5175
	if (err) {
5176
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
5177 5178
		goto out_nfserr;
	}
5179 5180
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&locku->lu_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
5181 5182
fput:
	fput(filp);
L
Linus Torvalds 已提交
5183
out:
5184
	nfsd4_bump_seqid(cstate, status);
5185 5186
	if (!cstate->replay_owner)
		nfs4_unlock_state();
5187 5188
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
5189 5190 5191
	return status;

out_nfserr:
5192
	status = nfserrno(err);
5193
	goto fput;
L
Linus Torvalds 已提交
5194 5195 5196 5197
}

/*
 * returns
5198 5199
 * 	true:  locks held by lockowner
 * 	false: no locks held by lockowner
L
Linus Torvalds 已提交
5200
 */
5201 5202
static bool
check_for_locks(struct nfs4_file *fp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
5203 5204
{
	struct file_lock **flpp;
5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215
	int status = false;
	struct file *filp = find_any_file(fp);
	struct inode *inode;

	if (!filp) {
		/* Any valid lock stateid should have some sort of access */
		WARN_ON_ONCE(1);
		return status;
	}

	inode = file_inode(filp);
L
Linus Torvalds 已提交
5216

5217
	spin_lock(&inode->i_lock);
L
Linus Torvalds 已提交
5218
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
5219
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
5220 5221
			status = true;
			break;
5222
		}
L
Linus Torvalds 已提交
5223
	}
5224
	spin_unlock(&inode->i_lock);
5225
	fput(filp);
L
Linus Torvalds 已提交
5226 5227 5228
	return status;
}

5229
__be32
5230 5231 5232
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
5233 5234
{
	clientid_t *clid = &rlockowner->rl_clientid;
5235
	struct nfs4_stateowner *sop = NULL, *tmp;
5236
	struct nfs4_lockowner *lo;
5237
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
5238
	struct xdr_netobj *owner = &rlockowner->rl_owner;
5239
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
5240
	__be32 status;
5241
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
5242 5243 5244 5245 5246 5247

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

	nfs4_lock_state();

5248
	status = lookup_clientid(clid, cstate, nn);
5249 5250 5251
	if (status)
		goto out;

5252
	status = nfserr_locks_held;
5253

5254 5255 5256
	/* Find the matching lock stateowner */
	list_for_each_entry(tmp, &nn->ownerstr_hashtbl[hashval], so_strhash) {
		if (tmp->so_is_open_owner)
5257
			continue;
5258 5259 5260
		if (same_owner_str(tmp, owner, clid)) {
			sop = tmp;
			break;
L
Linus Torvalds 已提交
5261
		}
5262
	}
5263 5264 5265 5266 5267 5268 5269 5270 5271 5272

	/* No matching owner found, maybe a replay? Just declare victory... */
	if (!sop) {
		status = nfs_ok;
		goto out;
	}

	lo = lockowner(sop);
	/* see if there are still any locks associated with it */
	list_for_each_entry(stp, &sop->so_stateids, st_perstateowner) {
5273
		if (check_for_locks(stp->st_stid.sc_file, lo))
5274
			goto out;
L
Linus Torvalds 已提交
5275
	}
5276 5277 5278

	status = nfs_ok;
	release_lockowner(lo);
L
Linus Torvalds 已提交
5279 5280 5281 5282 5283 5284
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
5285
alloc_reclaim(void)
L
Linus Torvalds 已提交
5286
{
N
NeilBrown 已提交
5287
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
5288 5289
}

5290
bool
5291
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
5292
{
5293
	struct nfs4_client_reclaim *crp;
5294

5295
	crp = nfsd4_find_reclaim_client(name, nn);
5296
	return (crp && crp->cr_clp);
5297 5298
}

L
Linus Torvalds 已提交
5299 5300 5301
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
5302
struct nfs4_client_reclaim *
5303
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
5304 5305
{
	unsigned int strhashval;
5306
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
5307

N
NeilBrown 已提交
5308 5309
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
5310 5311 5312
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
5313
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
5314
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
5315
		crp->cr_clp = NULL;
5316
		nn->reclaim_str_hashtbl_size++;
5317 5318
	}
	return crp;
L
Linus Torvalds 已提交
5319 5320
}

5321
void
5322
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
5323 5324 5325
{
	list_del(&crp->cr_strhash);
	kfree(crp);
5326
	nn->reclaim_str_hashtbl_size--;
5327 5328
}

5329
void
5330
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
5331 5332 5333 5334 5335
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
5336 5337
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
5338
			                struct nfs4_client_reclaim, cr_strhash);
5339
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
5340 5341
		}
	}
5342
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
5343 5344 5345 5346
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
5347
struct nfs4_client_reclaim *
5348
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
5349 5350 5351 5352
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

5353
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
5354

5355
	strhashval = clientstr_hashval(recdir);
5356
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
5357
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
5358 5359 5360 5361 5362 5363 5364 5365 5366
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
5367
__be32
5368 5369 5370
nfs4_check_open_reclaim(clientid_t *clid,
		struct nfsd4_compound_state *cstate,
		struct nfsd_net *nn)
L
Linus Torvalds 已提交
5371
{
5372
	__be32 status;
5373 5374

	/* find clientid in conf_id_hashtbl */
5375 5376
	status = lookup_clientid(clid, cstate, nn);
	if (status)
5377 5378
		return nfserr_reclaim_bad;

5379 5380 5381 5382
	if (nfsd4_client_record_check(cstate->clp))
		return nfserr_reclaim_bad;

	return nfs_ok;
L
Linus Torvalds 已提交
5383 5384
}

B
Bryan Schumaker 已提交
5385 5386
#ifdef CONFIG_NFSD_FAULT_INJECTION

5387 5388
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
5389 5390
	if (mark_client_expired(clp))
		return 0;
5391 5392 5393 5394
	expire_client(clp);
	return 1;
}

5395 5396 5397
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
5398
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
5399 5400 5401 5402 5403 5404 5405 5406
	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];
5407
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
5408 5409 5410
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

5411 5412
static u64 nfsd_foreach_client_lock(struct nfs4_client *clp, u64 max,
				    void (*func)(struct nfs4_ol_stateid *))
B
Bryan Schumaker 已提交
5413 5414 5415
{
	struct nfs4_openowner *oop;
	struct nfs4_ol_stateid *stp, *st_next;
5416
	struct nfs4_ol_stateid *lst, *lst_next;
B
Bryan Schumaker 已提交
5417 5418 5419
	u64 count = 0;

	list_for_each_entry(oop, &clp->cl_openowners, oo_perclient) {
5420 5421 5422 5423
		list_for_each_entry_safe(stp, st_next,
				&oop->oo_owner.so_stateids, st_perstateowner) {
			list_for_each_entry_safe(lst, lst_next,
					&stp->st_locks, st_locks) {
B
Bryan Schumaker 已提交
5424
				if (func)
5425
					func(lst);
B
Bryan Schumaker 已提交
5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436
				if (++count == max)
					return count;
			}
		}
	}

	return count;
}

u64 nfsd_forget_client_locks(struct nfs4_client *clp, u64 max)
{
5437
	return nfsd_foreach_client_lock(clp, max, release_lock_stateid);
B
Bryan Schumaker 已提交
5438 5439
}

5440 5441 5442 5443 5444 5445 5446
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;
}

5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466
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);
}

5467 5468 5469 5470 5471 5472 5473
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;
}

5474 5475 5476 5477 5478 5479
static u64 nfsd_find_all_delegations(struct nfs4_client *clp, u64 max,
				     struct list_head *victims)
{
	struct nfs4_delegation *dp, *next;
	u64 count = 0;

5480
	lockdep_assert_held(&state_lock);
5481
	list_for_each_entry_safe(dp, next, &clp->cl_delegations, dl_perclnt) {
5482 5483 5484 5485 5486 5487 5488 5489 5490 5491
		if (victims) {
			/*
			 * It's not safe to mess with delegations that have a
			 * non-zero dl_time. They might have already been broken
			 * and could be processed by the laundromat outside of
			 * the state_lock. Just leave them be.
			 */
			if (dp->dl_time != 0)
				continue;

5492 5493
			unhash_delegation_locked(dp);
			list_add(&dp->dl_recall_lru, victims);
5494
		}
5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506
		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;

5507
	spin_lock(&state_lock);
5508
	count = nfsd_find_all_delegations(clp, max, &victims);
5509
	spin_unlock(&state_lock);
5510

5511 5512
	list_for_each_entry_safe(dp, next, &victims, dl_recall_lru) {
		list_del_init(&dp->dl_recall_lru);
5513
		revoke_delegation(dp);
5514
	}
5515 5516 5517 5518 5519 5520

	return count;
}

u64 nfsd_recall_client_delegations(struct nfs4_client *clp, u64 max)
{
5521
	struct nfs4_delegation *dp;
5522 5523 5524
	LIST_HEAD(victims);
	u64 count;

5525
	spin_lock(&state_lock);
5526
	count = nfsd_find_all_delegations(clp, max, &victims);
5527 5528 5529 5530 5531
	while (!list_empty(&victims)) {
		dp = list_first_entry(&victims, struct nfs4_delegation,
					dl_recall_lru);
		list_del_init(&dp->dl_recall_lru);
		dp->dl_time = 0;
5532
		nfsd_break_one_deleg(dp);
5533
	}
5534
	spin_unlock(&state_lock);
5535 5536 5537 5538

	return count;
}

5539 5540 5541 5542
u64 nfsd_print_client_delegations(struct nfs4_client *clp, u64 max)
{
	u64 count = 0;

5543
	spin_lock(&state_lock);
5544
	count = nfsd_find_all_delegations(clp, max, NULL);
5545
	spin_unlock(&state_lock);
5546 5547 5548 5549 5550

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

5551
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
5552 5553
{
	struct nfs4_client *clp, *next;
5554
	u64 count = 0;
5555
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
5556

5557 5558 5559
	if (!nfsd_netns_ready(nn))
		return 0;

5560
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
5561 5562
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
5563 5564 5565
			break;
	}

5566 5567 5568
	return count;
}

5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583
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
5584 5585
#endif /* CONFIG_NFSD_FAULT_INJECTION */

5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606
/*
 * 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);
}

5607
static int nfs4_state_create_net(struct net *net)
5608 5609 5610 5611 5612 5613 5614
{
	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)
5615
		goto err;
5616 5617 5618 5619
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
5620 5621 5622 5623
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
5624 5625 5626 5627
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
5628

5629
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
5630
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
5631
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
5632
	}
5633 5634
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
5635 5636
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
5637
	nn->conf_name_tree = RB_ROOT;
5638
	nn->unconf_name_tree = RB_ROOT;
5639
	INIT_LIST_HEAD(&nn->client_lru);
5640
	INIT_LIST_HEAD(&nn->close_lru);
5641
	INIT_LIST_HEAD(&nn->del_recall_lru);
5642
	spin_lock_init(&nn->client_lock);
5643

5644
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
5645
	get_net(net);
5646

5647
	return 0;
5648

5649
err_sessionid:
5650
	kfree(nn->ownerstr_hashtbl);
5651 5652
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
5653 5654
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
5655 5656
err:
	return -ENOMEM;
5657 5658 5659
}

static void
5660
nfs4_state_destroy_net(struct net *net)
5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

	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);
		}
	}
5672

5673 5674 5675 5676 5677
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
		while (!list_empty(&nn->unconf_id_hashtbl[i])) {
			clp = list_entry(nn->unconf_id_hashtbl[i].next, struct nfs4_client, cl_idhash);
			destroy_client(clp);
		}
5678 5679
	}

5680
	kfree(nn->sessionid_hashtbl);
5681
	kfree(nn->ownerstr_hashtbl);
5682
	kfree(nn->unconf_id_hashtbl);
5683
	kfree(nn->conf_id_hashtbl);
5684
	put_net(net);
5685 5686
}

5687
int
5688
nfs4_state_start_net(struct net *net)
5689
{
5690
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
5691 5692
	int ret;

5693
	ret = nfs4_state_create_net(net);
5694 5695
	if (ret)
		return ret;
5696
	nfsd4_client_tracking_init(net);
5697
	nn->boot_time = get_seconds();
5698
	locks_start_grace(net, &nn->nfsd4_manager);
5699
	nn->grace_ended = false;
5700
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
5701 5702
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
5703 5704 5705 5706 5707 5708 5709 5710 5711 5712
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

5713
	ret = set_callback_cred();
5714 5715
	if (ret)
		return -ENOMEM;
5716
	laundry_wq = create_singlethread_workqueue("nfsd4");
5717 5718 5719 5720
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
5721 5722 5723
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
5724

5725
	set_max_delegations();
5726

5727
	return 0;
5728

5729 5730
out_free_laundry:
	destroy_workqueue(laundry_wq);
5731
out_recovery:
5732
	return ret;
L
Linus Torvalds 已提交
5733 5734
}

5735
void
5736
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
5737 5738 5739
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
5740
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
5741

5742 5743
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
5744

5745
	nfs4_lock_state();
L
Linus Torvalds 已提交
5746
	INIT_LIST_HEAD(&reaplist);
5747
	spin_lock(&state_lock);
5748
	list_for_each_safe(pos, next, &nn->del_recall_lru) {
L
Linus Torvalds 已提交
5749
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
5750 5751
		unhash_delegation_locked(dp);
		list_add(&dp->dl_recall_lru, &reaplist);
L
Linus Torvalds 已提交
5752
	}
5753
	spin_unlock(&state_lock);
L
Linus Torvalds 已提交
5754 5755
	list_for_each_safe(pos, next, &reaplist) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
5756
		list_del_init(&dp->dl_recall_lru);
5757
		nfs4_put_stid(&dp->dl_stid);
L
Linus Torvalds 已提交
5758 5759
	}

5760
	nfsd4_client_tracking_exit(net);
5761
	nfs4_state_destroy_net(net);
5762
	nfs4_unlock_state();
L
Linus Torvalds 已提交
5763 5764 5765 5766 5767
}

void
nfs4_state_shutdown(void)
{
5768
	destroy_workqueue(laundry_wq);
5769
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
5770
}
5771 5772 5773 5774

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5775 5776
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
5777 5778 5779 5780 5781
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5782 5783 5784 5785 5786 5787 5788 5789 5790 5791
	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);
5792 5793
}

5794 5795 5796
/*
 * functions to set current state id
 */
5797 5798 5799 5800 5801 5802
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5803 5804 5805 5806 5807 5808
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823
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
 */
5824

5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836
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);
}

5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848
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);
}

5849 5850 5851 5852 5853 5854 5855
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5856
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5857
{
5858
	get_stateid(cstate, &locku->lu_stateid);
5859
}
5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871

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);
}