nfs4state.c 132.2 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/clnt.h>
44
#include "xdr4.h"
45
#include "vfs.h"
46
#include "current_stateid.h"
47
#include "fault_inject.h"
L
Linus Torvalds 已提交
48

49 50
#include "netns.h"

L
Linus Torvalds 已提交
51 52 53
#define NFSDDBG_FACILITY                NFSDDBG_PROC

/* Globals */
54
time_t nfsd4_lease = 90;     /* default lease time */
55
time_t nfsd4_grace = 90;
56 57 58 59 60 61 62 63 64

#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 */
};
65 66 67
static const stateid_t currentstateid = {
	.si_generation = 1,
};
68

A
Andy Adamson 已提交
69
static u64 current_sessionid = 1;
70

71 72
#define ZERO_STATEID(stateid) (!memcmp((stateid), &zero_stateid, sizeof(stateid_t)))
#define ONE_STATEID(stateid)  (!memcmp((stateid), &one_stateid, sizeof(stateid_t)))
73
#define CURRENT_STATEID(stateid) (!memcmp((stateid), &currentstateid, sizeof(stateid_t)))
L
Linus Torvalds 已提交
74 75

/* forward declarations */
76
static int check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner);
L
Linus Torvalds 已提交
77

78 79 80
/* Locking: */

/* Currently used for almost all code touching nfsv4 state: */
I
Ingo Molnar 已提交
81
static DEFINE_MUTEX(client_mutex);
L
Linus Torvalds 已提交
82

83 84 85 86 87 88 89
/*
 * Currently used for the del_recall_lru and file hash table.  In an
 * effort to decrease the scope of the client_mutex, this spinlock may
 * eventually cover more:
 */
static DEFINE_SPINLOCK(recall_lock);

90 91
static struct kmem_cache *openowner_slab = NULL;
static struct kmem_cache *lockowner_slab = NULL;
92 93 94
static struct kmem_cache *file_slab = NULL;
static struct kmem_cache *stateid_slab = NULL;
static struct kmem_cache *deleg_slab = NULL;
N
NeilBrown 已提交
95

L
Linus Torvalds 已提交
96 97 98
void
nfs4_lock_state(void)
{
I
Ingo Molnar 已提交
99
	mutex_lock(&client_mutex);
L
Linus Torvalds 已提交
100 101
}

102 103 104 105 106 107 108 109 110 111 112 113 114
static void free_session(struct kref *);

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

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

L
Linus Torvalds 已提交
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
}

static inline u32
opaque_hashval(const void *ptr, int nbytes)
{
	unsigned char *cptr = (unsigned char *) ptr;

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

static struct list_head del_recall_lru;

136 137 138 139 140
static void nfsd4_free_file(struct nfs4_file *f)
{
	kmem_cache_free(file_slab, f);
}

141 142 143
static inline void
put_nfs4_file(struct nfs4_file *fi)
{
144 145 146 147
	if (atomic_dec_and_lock(&fi->fi_ref, &recall_lock)) {
		list_del(&fi->fi_hash);
		spin_unlock(&recall_lock);
		iput(fi->fi_inode);
148
		nfsd4_free_file(fi);
149
	}
150 151 152 153 154
}

static inline void
get_nfs4_file(struct nfs4_file *fi)
{
155
	atomic_inc(&fi->fi_ref);
156 157
}

158
static int num_delegations;
159
unsigned int max_delegations;
160 161 162 163 164

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

165 166 167 168
/* 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)
169

170
static unsigned int ownerstr_hashval(u32 clientid, struct xdr_netobj *ownername)
171 172 173 174 175
{
	unsigned int ret;

	ret = opaque_hashval(ownername->data, ownername->len);
	ret += clientid;
176
	return ret & OWNER_HASH_MASK;
177
}
178 179 180 181

/* hash table for nfs4_file */
#define FILE_HASH_BITS                   8
#define FILE_HASH_SIZE                  (1 << FILE_HASH_BITS)
S
Shan Wei 已提交
182

183 184 185 186 187 188
static unsigned int file_hashval(struct inode *ino)
{
	/* XXX: why are we hashing on inode pointer, anyway? */
	return hash_ptr(ino, FILE_HASH_BITS);
}

189 190
static struct list_head file_hashtbl[FILE_HASH_SIZE];

191
static void __nfs4_file_get_access(struct nfs4_file *fp, int oflag)
192
{
193
	WARN_ON_ONCE(!(fp->fi_fds[oflag] || fp->fi_fds[O_RDWR]));
194 195 196
	atomic_inc(&fp->fi_access[oflag]);
}

197 198 199 200 201 202 203 204 205 206
static void nfs4_file_get_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_get_access(fp, O_RDONLY);
		__nfs4_file_get_access(fp, O_WRONLY);
	} else
		__nfs4_file_get_access(fp, oflag);
}

static void nfs4_file_put_fd(struct nfs4_file *fp, int oflag)
207 208 209 210 211 212 213
{
	if (fp->fi_fds[oflag]) {
		fput(fp->fi_fds[oflag]);
		fp->fi_fds[oflag] = NULL;
	}
}

214
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
215 216 217
{
	if (atomic_dec_and_test(&fp->fi_access[oflag])) {
		nfs4_file_put_fd(fp, oflag);
218 219 220 221 222 223 224 225
		/*
		 * It's also safe to get rid of the RDWR open *if*
		 * we no longer have need of the other kind of access
		 * or if we already have the other kind of open:
		 */
		if (fp->fi_fds[1-oflag]
			|| atomic_read(&fp->fi_access[1 - oflag]) == 0)
			nfs4_file_put_fd(fp, O_RDWR);
226 227 228
	}
}

229 230 231 232 233 234 235 236 237
static void nfs4_file_put_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_put_access(fp, O_RDONLY);
		__nfs4_file_put_access(fp, O_WRONLY);
	} else
		__nfs4_file_put_access(fp, oflag);
}

J
J. Bruce Fields 已提交
238
static inline int get_new_stid(struct nfs4_stid *stid)
239
{
J
J. Bruce Fields 已提交
240
	static int min_stateid = 0;
241
	struct idr *stateids = &stid->sc_client->cl_stateids;
J
J. Bruce Fields 已提交
242 243 244
	int new_stid;
	int error;

245
	error = idr_get_new_above(stateids, stid, min_stateid, &new_stid);
J
J. Bruce Fields 已提交
246
	/*
247 248 249 250
	 * Note: the necessary preallocation was done in
	 * nfs4_alloc_stateid().  The idr code caps the number of
	 * preallocations that can exist at a time, but the state lock
	 * prevents anyone from using ours before we get here:
J
J. Bruce Fields 已提交
251
	 */
252
	WARN_ON_ONCE(error);
J
J. Bruce Fields 已提交
253 254 255 256 257 258 259 260 261
	/*
	 * 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):
	 */
262

J
J. Bruce Fields 已提交
263 264 265 266
	min_stateid = new_stid+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return new_stid;
267 268
}

269
static void init_stid(struct nfs4_stid *stid, struct nfs4_client *cl, unsigned char type)
270 271
{
	stateid_t *s = &stid->sc_stateid;
J
J. Bruce Fields 已提交
272
	int new_id;
273 274 275 276

	stid->sc_type = type;
	stid->sc_client = cl;
	s->si_opaque.so_clid = cl->cl_clientid;
J
J. Bruce Fields 已提交
277 278
	new_id = get_new_stid(stid);
	s->si_opaque.so_id = (u32)new_id;
279 280
	/* Will be incremented before return to client: */
	s->si_generation = 0;
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
}

static struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl, struct kmem_cache *slab)
{
	struct idr *stateids = &cl->cl_stateids;

	if (!idr_pre_get(stateids, GFP_KERNEL))
		return NULL;
	/*
	 * Note: if we fail here (or any time between now and the time
	 * we actually get the new idr), we won't need to undo the idr
	 * preallocation, since the idr code caps the number of
	 * preallocated entries.
	 */
	return kmem_cache_alloc(slab, GFP_KERNEL);
296 297
}

298 299 300 301 302
static struct nfs4_ol_stateid * nfs4_alloc_stateid(struct nfs4_client *clp)
{
	return openlockstateid(nfs4_alloc_stid(clp, stateid_slab));
}

L
Linus Torvalds 已提交
303
static struct nfs4_delegation *
304
alloc_init_deleg(struct nfs4_client *clp, struct nfs4_ol_stateid *stp, struct svc_fh *current_fh, u32 type)
L
Linus Torvalds 已提交
305 306 307 308 309
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
310 311 312 313 314 315 316
	/*
	 * Major work on the lease subsystem (for example, to support
	 * calbacks on stat) will be required before we can support
	 * write delegations properly.
	 */
	if (type != NFS4_OPEN_DELEGATE_READ)
		return NULL;
317 318
	if (fp->fi_had_conflict)
		return NULL;
319
	if (num_delegations > max_delegations)
320
		return NULL;
321
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
322
	if (dp == NULL)
L
Linus Torvalds 已提交
323
		return dp;
324
	init_stid(&dp->dl_stid, clp, NFS4_DELEG_STID);
325 326
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
327 328
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
329 330
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
331
	num_delegations++;
332 333
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
334
	INIT_LIST_HEAD(&dp->dl_recall_lru);
335
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
336 337
	dp->dl_file = fp;
	dp->dl_type = type;
338
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
339 340
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
341
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
342 343 344 345 346 347 348 349
	return dp;
}

void
nfs4_put_delegation(struct nfs4_delegation *dp)
{
	if (atomic_dec_and_test(&dp->dl_count)) {
		dprintk("NFSD: freeing dp %p\n",dp);
350
		put_nfs4_file(dp->dl_file);
N
NeilBrown 已提交
351
		kmem_cache_free(deleg_slab, dp);
352
		num_delegations--;
L
Linus Torvalds 已提交
353 354 355
	}
}

356
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
357
{
358 359 360
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
361
		fput(fp->fi_deleg_file);
362 363
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
364 365
}

J
J. Bruce Fields 已提交
366 367
static void unhash_stid(struct nfs4_stid *s)
{
368 369 370
	struct idr *stateids = &s->sc_client->cl_stateids;

	idr_remove(stateids, s->sc_stateid.si_opaque.so_id);
J
J. Bruce Fields 已提交
371 372
}

L
Linus Torvalds 已提交
373 374 375 376
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
377
	unhash_stid(&dp->dl_stid);
378
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
379
	spin_lock(&recall_lock);
380
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
381 382
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
383
	nfs4_put_deleg_lease(dp->dl_file);
L
Linus Torvalds 已提交
384 385 386 387 388 389 390
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

391 392 393 394 395 396 397 398 399 400
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;
}

401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
/*
 * 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.
 */
419 420
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
421
	int i;
422
	unsigned int access = 0;
423 424 425

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
426
			access |= i;
427
	}
428
	return access;
429 430
}

431
static bool
432
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
433 434
	unsigned int access, deny;

435 436
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
437
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
438 439
		return false;
	return true;
440 441
}

442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
/* set share access for a given stateid */
static inline void
set_access(u32 access, struct nfs4_ol_stateid *stp)
{
	__set_bit(access, &stp->st_access_bmap);
}

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

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

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
/* set share deny for a given stateid */
static inline void
set_deny(u32 access, struct nfs4_ol_stateid *stp)
{
	__set_bit(access, &stp->st_deny_bmap);
}

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

/* test whether a given stateid is denying specific access */
static inline bool
test_deny(u32 access, struct nfs4_ol_stateid *stp)
{
	return test_bit(access, &stp->st_deny_bmap);
482 483 484 485
}

static int nfs4_access_to_omode(u32 access)
{
486
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
487 488 489 490 491 492 493
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
494 495
	WARN_ON_ONCE(1);
	return O_RDONLY;
496 497
}

498 499 500 501 502 503 504 505 506 507 508 509 510 511
/* release all access and file references for a given stateid */
static void
release_all_access(struct nfs4_ol_stateid *stp)
{
	int i;

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

512
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
513 514 515 516 517
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

518
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
519
{
520
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
521
	put_nfs4_file(stp->st_file);
522 523 524
	stp->st_file = NULL;
}

525
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
526
{
527 528 529
	kmem_cache_free(stateid_slab, stp);
}

530
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
531 532 533 534
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
535
	unhash_stid(&stp->st_stid);
536 537
	file = find_any_file(stp->st_file);
	if (file)
538
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
539
	close_generic_stateid(stp);
540 541 542
	free_generic_stateid(stp);
}

543
static void unhash_lockowner(struct nfs4_lockowner *lo)
544
{
545
	struct nfs4_ol_stateid *stp;
546

547 548
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
549
	list_del(&lo->lo_owner_ino_hash);
550 551
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
552
				struct nfs4_ol_stateid, st_perstateowner);
553 554 555 556
		release_lock_stateid(stp);
	}
}

557
static void release_lockowner(struct nfs4_lockowner *lo)
558
{
559 560
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
561 562 563
}

static void
564
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
565
{
566
	struct nfs4_lockowner *lo;
567 568

	while (!list_empty(&open_stp->st_lockowners)) {
569 570 571
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
572 573 574
	}
}

575
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
576 577 578
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
579 580 581 582 583 584
	close_generic_stateid(stp);
}

static void release_open_stateid(struct nfs4_ol_stateid *stp)
{
	unhash_open_stateid(stp);
J
J. Bruce Fields 已提交
585
	unhash_stid(&stp->st_stid);
586 587 588
	free_generic_stateid(stp);
}

589
static void unhash_openowner(struct nfs4_openowner *oo)
590
{
591
	struct nfs4_ol_stateid *stp;
592

593 594 595 596
	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,
597
				struct nfs4_ol_stateid, st_perstateowner);
598
		release_open_stateid(stp);
599 600 601
	}
}

602 603 604 605 606
static void release_last_closed_stateid(struct nfs4_openowner *oo)
{
	struct nfs4_ol_stateid *s = oo->oo_last_closed_stid;

	if (s) {
J
J. Bruce Fields 已提交
607
		unhash_stid(&s->st_stid);
608 609 610 611 612
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

613
static void release_openowner(struct nfs4_openowner *oo)
614
{
615 616
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
617
	release_last_closed_stateid(oo);
618
	nfs4_free_openowner(oo);
619 620
}

M
Marc Eshel 已提交
621 622 623 624 625 626 627 628
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

629
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
630 631 632 633 634 635
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]);
}
636 637 638 639 640 641 642
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

M
Marc Eshel 已提交
643

A
Andy Adamson 已提交
644 645 646 647 648 649 650 651 652 653 654 655 656
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;
}

/*
657 658 659 660 661 662 663 664 665 666 667 668 669
 * 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)

670 671 672 673 674 675 676 677 678
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]);
}

679
/*
680 681 682 683 684 685 686 687
 * We don't actually need to cache the rpc and session headers, so we
 * can allocate a little less for each slot:
 */
static inline int slot_bytes(struct nfsd4_channel_attrs *ca)
{
	return ca->maxresp_cached - NFSD_MIN_HDR_SEQ_SZ;
}

688
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
689
{
690 691
	size -= NFSD_MIN_HDR_SEQ_SZ; /* We don't cache the rpc header */
	size = min_t(u32, size, NFSD_SLOT_CACHE_SIZE);
A
Andy Adamson 已提交
692

693 694
	return size;
}
A
Andy Adamson 已提交
695

696 697
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
698
 * re-negotiate active sessions and reduce their slot usage to make
699
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
700
 */
701
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
702
{
703
	int avail;
A
Andy Adamson 已提交
704

705
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
706

707 708 709 710 711 712
	spin_lock(&nfsd_drc_lock);
	avail = min_t(int, NFSD_MAX_MEM_PER_SESSION,
			nfsd_drc_max_mem - nfsd_drc_mem_used);
	num = min_t(int, num, avail / slotsize);
	nfsd_drc_mem_used += num * slotsize;
	spin_unlock(&nfsd_drc_lock);
A
Andy Adamson 已提交
713

714 715
	return num;
}
A
Andy Adamson 已提交
716

717 718
static void nfsd4_put_drc_mem(int slotsize, int num)
{
719
	spin_lock(&nfsd_drc_lock);
720
	nfsd_drc_mem_used -= slotsize * num;
721
	spin_unlock(&nfsd_drc_lock);
722
}
A
Andy Adamson 已提交
723

724
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
725 726 727
{
	struct nfsd4_session *new;
	int mem, i;
728

729 730 731
	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
732

733 734 735
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
736
	/* allocate each struct nfsd4_slot and data cache in one piece */
737 738 739 740
	for (i = 0; i < numslots; i++) {
		mem = sizeof(struct nfsd4_slot) + slotsize;
		new->se_slots[i] = kzalloc(mem, GFP_KERNEL);
		if (!new->se_slots[i])
741 742
			goto out_free;
	}
743 744 745 746 747 748
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
749 750
}

751
static void init_forechannel_attrs(struct nfsd4_channel_attrs *new, struct nfsd4_channel_attrs *req, int numslots, int slotsize)
A
Andy Adamson 已提交
752
{
753
	u32 maxrpc = nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
754

755
	new->maxreqs = numslots;
756 757
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
758 759 760 761
	new->maxreq_sz = min_t(u32, req->maxreq_sz, maxrpc);
	new->maxresp_sz = min_t(u32, req->maxresp_sz, maxrpc);
	new->maxops = min_t(u32, req->maxops, NFSD_MAX_OPS_PER_COMPOUND);
}
A
Andy Adamson 已提交
762

763 764 765 766 767
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
768

769 770 771 772
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
773

774 775 776 777 778 779
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
	spin_unlock(&clp->cl_lock);
780
	nfsd4_probe_callback(clp);
781
}
A
Andy Adamson 已提交
782

783
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
784 785
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
786

787 788
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
789
		return NULL;
790 791
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
792
	conn->cn_flags = flags;
793 794 795
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
796

797 798 799 800
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
801 802
}

803
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
804
{
805
	struct nfs4_client *clp = ses->se_client;
806

807
	spin_lock(&clp->cl_lock);
808
	__nfsd4_hash_conn(conn, ses);
809
	spin_unlock(&clp->cl_lock);
810 811
}

812
static int nfsd4_register_conn(struct nfsd4_conn *conn)
813
{
814
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
815
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
816 817
}

818
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
819
{
820
	int ret;
A
Andy Adamson 已提交
821

822
	nfsd4_hash_conn(conn, ses);
823 824 825 826
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
827
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
828 829 830
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
831
}
A
Andy Adamson 已提交
832

833
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
834 835 836
{
	u32 dir = NFS4_CDFC4_FORE;

837
	if (cses->flags & SESSION4_BACK_CHAN)
838
		dir |= NFS4_CDFC4_BACK;
839
	return alloc_conn(rqstp, dir);
840 841 842
}

/* must be called under client_lock */
843
static void nfsd4_del_conns(struct nfsd4_session *s)
844
{
845 846
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
847

848 849 850 851 852
	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);
853

854 855
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
856

857 858 859
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
860
}
A
Andy Adamson 已提交
861

862 863 864 865 866 867 868
static void __free_session(struct nfsd4_session *ses)
{
	nfsd4_put_drc_mem(slot_bytes(&ses->se_fchannel), ses->se_fchannel.maxreqs);
	free_session_slots(ses);
	kfree(ses);
}

869
static void free_session(struct kref *kref)
870 871
{
	struct nfsd4_session *ses;
872
	struct nfsd_net *nn;
873 874

	ses = container_of(kref, struct nfsd4_session, se_ref);
875 876 877
	nn = net_generic(ses->se_client->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
878
	nfsd4_del_conns(ses);
879
	__free_session(ses);
880 881
}

882 883
void nfsd4_put_session(struct nfsd4_session *ses)
{
884 885 886
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
887
	nfsd4_put_session_locked(ses);
888
	spin_unlock(&nn->client_lock);
889 890
}

891
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan)
892 893 894 895 896 897 898 899 900 901 902 903
{
	struct nfsd4_session *new;
	int numslots, slotsize;
	/*
	 * Note decreasing slot size below client's request may
	 * make it difficult for client to function correctly, whereas
	 * decreasing the number of slots will (just?) affect
	 * performance.  When short on memory we therefore prefer to
	 * decrease number of slots instead of their size.
	 */
	slotsize = nfsd4_sanitize_slot_size(fchan->maxresp_cached);
	numslots = nfsd4_get_drc_mem(slotsize, fchan->maxreqs);
904 905
	if (numslots < 1)
		return NULL;
906

907
	new = __alloc_session(slotsize, numslots);
908 909
	if (!new) {
		nfsd4_put_drc_mem(slotsize, fchan->maxreqs);
910
		return NULL;
911
	}
912
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize);
913 914
	return new;
}
915

916
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
917 918
{
	int idx;
919
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
920

A
Andy Adamson 已提交
921 922 923
	new->se_client = clp;
	gen_sessionid(new);

924 925
	INIT_LIST_HEAD(&new->se_conns);

926
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
927
	new->se_flags = cses->flags;
928
	new->se_cb_prog = cses->callback_prog;
929
	new->se_cb_sec = cses->cb_sec;
A
Andy Adamson 已提交
930
	kref_init(&new->se_ref);
931
	idx = hash_sessionid(&new->se_sessionid);
932
	spin_lock(&nn->client_lock);
933
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
934
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
935
	list_add(&new->se_perclnt, &clp->cl_sessions);
936
	spin_unlock(&clp->cl_lock);
937
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
938

939
	if (cses->flags & SESSION4_BACK_CHAN) {
940
		struct sockaddr *sa = svc_addr(rqstp);
941 942 943 944 945 946 947
		/*
		 * 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:
		 */
948 949 950
		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
951 952
}

953
/* caller must hold client_lock */
M
Marc Eshel 已提交
954
static struct nfsd4_session *
955
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
956 957 958
{
	struct nfsd4_session *elem;
	int idx;
959
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
960 961 962 963

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
964
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
965 966 967 968 969 970 971 972 973 974
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

975
/* caller must hold client_lock */
A
Andy Adamson 已提交
976
static void
M
Marc Eshel 已提交
977
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
978 979
{
	list_del(&ses->se_hash);
980
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
981
	list_del(&ses->se_perclnt);
982
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
983 984
}

985
/* must be called under the client_lock */
L
Linus Torvalds 已提交
986
static inline void
987
renew_client_locked(struct nfs4_client *clp)
L
Linus Torvalds 已提交
988
{
989 990
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

B
Benny Halevy 已提交
991
	if (is_client_expired(clp)) {
992 993
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
B
Benny Halevy 已提交
994 995 996 997 998 999
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

L
Linus Torvalds 已提交
1000 1001 1002
	dprintk("renewing client (clientid %08x/%08x)\n", 
			clp->cl_clientid.cl_boot, 
			clp->cl_clientid.cl_id);
1003
	list_move_tail(&clp->cl_lru, &nn->client_lru);
L
Linus Torvalds 已提交
1004 1005 1006
	clp->cl_time = get_seconds();
}

1007 1008 1009
static inline void
renew_client(struct nfs4_client *clp)
{
1010 1011 1012
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
1013
	renew_client_locked(clp);
1014
	spin_unlock(&nn->client_lock);
1015 1016
}

L
Linus Torvalds 已提交
1017 1018
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1019
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1020
{
1021
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1022
		return 0;
A
Andy Adamson 已提交
1023
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1024
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1025 1026 1027 1028 1029 1030 1031 1032
	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.
 */
1033
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1034 1035 1036
{
	struct nfs4_client *clp;

1037 1038 1039
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1040
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1041 1042 1043
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1044
	}
1045
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1046 1047 1048 1049 1050 1051
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1052 1053 1054
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
1055 1056 1057 1058 1059
	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);
1060
		nfsd4_put_session_locked(ses);
1061
	}
1062
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1063 1064 1065 1066
	kfree(clp->cl_name.data);
	kfree(clp);
}

1067 1068 1069 1070
void
release_session_client(struct nfsd4_session *session)
{
	struct nfs4_client *clp = session->se_client;
1071
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
1072

1073
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
1074 1075 1076 1077 1078 1079
		return;
	if (is_client_expired(clp)) {
		free_client(clp);
		session->se_client = NULL;
	} else
		renew_client_locked(clp);
1080
	spin_unlock(&nn->client_lock);
1081 1082
}

B
Benny Halevy 已提交
1083 1084 1085 1086
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1087 1088
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1089
	mark_client_expired(clp);
B
Benny Halevy 已提交
1090
	list_del(&clp->cl_lru);
1091
	spin_lock(&clp->cl_lock);
1092 1093
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1094
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1095 1096
}

L
Linus Torvalds 已提交
1097
static void
1098
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1099
{
1100
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1101 1102
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1103
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1104 1105 1106

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1107 1108 1109
	while (!list_empty(&clp->cl_delegations)) {
		dp = list_entry(clp->cl_delegations.next, struct nfs4_delegation, dl_perclnt);
		list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
1110 1111 1112 1113 1114 1115 1116
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	while (!list_empty(&reaplist)) {
		dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}
1117
	while (!list_empty(&clp->cl_openowners)) {
1118 1119
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1120
	}
1121
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1122 1123
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1124
	list_del(&clp->cl_idhash);
1125
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1126
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1127
	else
1128
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1129
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1130
	unhash_client_locked(clp);
1131 1132
	if (atomic_read(&clp->cl_refcount) == 0)
		free_client(clp);
1133
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1134 1135
}

1136 1137 1138 1139 1140 1141
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1142 1143 1144 1145
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
1146 1147
}

1148 1149
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1150 1151 1152 1153
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1154
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1155
{
1156 1157 1158 1159 1160 1161 1162
	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;
1163
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1164 1165 1166 1167
	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);
1168
	return 0;
L
Linus Torvalds 已提交
1169 1170
}

1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
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);
}

1182
static int same_name(const char *n1, const char *n2)
1183
{
N
NeilBrown 已提交
1184
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1185 1186 1187
}

static int
1188 1189 1190
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1191 1192 1193
}

static int
1194 1195 1196
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
1197 1198
}

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
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++)
		if (GROUP_AT(g1, i) != GROUP_AT(g2, i))
			return false;
	return true;
}

1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226
/*
 * 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);
}


1227
static bool
1228 1229
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1230
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1231
		|| (cr1->cr_uid != cr2->cr_uid)
1232 1233 1234 1235 1236 1237 1238
		|| (cr1->cr_gid != cr2->cr_gid)
		|| !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;
1239
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1240 1241
}

1242
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1243 1244 1245
{
	static u32 current_clientid = 1;

1246
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1247 1248 1249
	clp->cl_clientid.cl_id = current_clientid++; 
}

1250 1251
static void gen_confirm(struct nfs4_client *clp)
{
1252
	__be32 verf[2];
1253
	static u32 i;
L
Linus Torvalds 已提交
1254

1255 1256 1257
	verf[0] = (__be32)get_seconds();
	verf[1] = (__be32)i++;
	memcpy(clp->cl_confirm.data, verf, sizeof(clp->cl_confirm.data));
L
Linus Torvalds 已提交
1258 1259
}

1260
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1261
{
1262
	return idr_find(&cl->cl_stateids, t->si_opaque.so_id);
J
J. Bruce Fields 已提交
1263 1264
}

1265
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1266 1267
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1268

1269
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1270 1271
	if (!s)
		return NULL;
1272
	if (typemask & s->sc_type)
1273 1274 1275 1276
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1277
static struct nfs4_client *create_client(struct xdr_netobj name,
1278 1279 1280 1281
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1282
	int ret;
1283
	struct net *net = SVC_NET(rqstp);
1284
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1285 1286 1287 1288 1289

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

1290
	INIT_LIST_HEAD(&clp->cl_sessions);
1291 1292
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1293
		spin_lock(&nn->client_lock);
1294
		free_client(clp);
1295
		spin_unlock(&nn->client_lock);
1296
		return NULL;
1297
	}
1298
	idr_init(&clp->cl_stateids);
1299
	atomic_set(&clp->cl_refcount, 0);
1300
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1301 1302 1303 1304
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1305
	INIT_LIST_HEAD(&clp->cl_callbacks);
1306
	spin_lock_init(&clp->cl_lock);
1307
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1308
	clp->cl_time = get_seconds();
1309 1310 1311 1312 1313
	clear_bit(0, &clp->cl_cb_slot_busy);
	rpc_init_wait_queue(&clp->cl_cb_waitq, "Backchannel slot table");
	copy_verf(clp, verf);
	rpc_copy_addr((struct sockaddr *) &clp->cl_addr, sa);
	gen_confirm(clp);
1314
	clp->cl_cb_session = NULL;
1315
	clp->net = net;
1316 1317 1318
	return clp;
}

1319
static void
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
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
1361 1362
{
	unsigned int idhashval;
1363
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1364

1365
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1366
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1367
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1368
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1369
	renew_client(clp);
L
Linus Torvalds 已提交
1370 1371
}

1372
static void
L
Linus Torvalds 已提交
1373 1374 1375
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1376
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1377 1378

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1379
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1380
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1381
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1382
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1383 1384 1385 1386
	renew_client(clp);
}

static struct nfs4_client *
1387
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1388 1389 1390 1391
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1392
	list_for_each_entry(clp, &nn->conf_id_hashtbl[idhashval], cl_idhash) {
1393
		if (same_clid(&clp->cl_clientid, clid)) {
1394 1395
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1396
			renew_client(clp);
L
Linus Torvalds 已提交
1397
			return clp;
1398
		}
L
Linus Torvalds 已提交
1399 1400 1401 1402 1403
	}
	return NULL;
}

static struct nfs4_client *
1404
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1405 1406 1407 1408
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1409
	list_for_each_entry(clp, &nn->unconf_id_hashtbl[idhashval], cl_idhash) {
1410 1411 1412
		if (same_clid(&clp->cl_clientid, clid)) {
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
L
Linus Torvalds 已提交
1413
			return clp;
1414
		}
L
Linus Torvalds 已提交
1415 1416 1417 1418
	}
	return NULL;
}

1419
static bool clp_used_exchangeid(struct nfs4_client *clp)
1420
{
1421
	return clp->cl_exchange_flags != 0;
1422
} 
1423

1424
static struct nfs4_client *
1425
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1426
{
1427
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1428 1429 1430
}

static struct nfs4_client *
1431
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1432
{
1433
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1434 1435
}

1436
static void
1437
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1438
{
1439
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1440 1441
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
	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
1452 1453
		goto out_err;

1454
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1455
					    se->se_callback_addr_len,
1456 1457
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1458

1459
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1460
		goto out_err;
1461

1462 1463
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1464

1465 1466
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1467
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1468 1469
	return;
out_err:
1470 1471
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1472
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1473 1474 1475 1476 1477 1478
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1479
/*
1480
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1481 1482 1483 1484
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1485 1486
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1487

1488
	dprintk("--> %s slot %p\n", __func__, slot);
1489

1490 1491
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1492

1493
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1494
	if (nfsd4_not_cached(resp)) {
1495
		slot->sl_datalen = 0;
1496
		return;
1497
	}
1498 1499 1500 1501 1502 1503 1504
	slot->sl_datalen = (char *)resp->p - (char *)resp->cstate.datap;
	base = (char *)resp->cstate.datap -
					(char *)resp->xbuf->head[0].iov_base;
	if (read_bytes_from_xdr_buf(resp->xbuf, base, slot->sl_data,
				    slot->sl_datalen))
		WARN("%s: sessions DRC could not cache compound\n", __func__);
	return;
1505 1506 1507
}

/*
1508 1509 1510 1511
 * 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.
1512 1513
 *
 */
1514 1515 1516
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1517
{
1518 1519
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1520

1521 1522 1523
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1524

1525
	/* Return nfserr_retry_uncached_rep in next operation. */
1526
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1527 1528 1529
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1530
	}
1531
	return op->status;
1532 1533 1534
}

/*
1535 1536
 * The sequence operation is not cached because we can use the slot and
 * session values.
1537 1538
 */
__be32
1539 1540
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1541
{
1542
	struct nfsd4_slot *slot = resp->cstate.slot;
1543 1544
	__be32 status;

1545
	dprintk("--> %s slot %p\n", __func__, slot);
1546

1547 1548 1549 1550
	/* Either returns 0 or nfserr_retry_uncached */
	status = nfsd4_enc_sequence_replay(resp->rqstp->rq_argp, resp);
	if (status == nfserr_retry_uncached_rep)
		return status;
1551

1552 1553
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1554

1555 1556 1557
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1558 1559 1560 1561

	return status;
}

A
Andy Adamson 已提交
1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
/*
 * 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;
}

1578 1579 1580 1581 1582 1583 1584 1585
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.
	 */
1586 1587 1588
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1589 1590
}

A
Andy Adamson 已提交
1591 1592 1593 1594 1595
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1596
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1597
	__be32 status;
1598
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1599
	nfs4_verifier		verf = exid->verifier;
1600
	struct sockaddr		*sa = svc_addr(rqstp);
1601
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1602
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1603

1604
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1605
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1606
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1607
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1608
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1609

1610
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1611 1612 1613 1614 1615 1616
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1617 1618
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1619 1620 1621 1622 1623
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1624
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1625
	nfs4_lock_state();
1626
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1627
	if (conf) {
1628 1629 1630
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1631 1632
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1633
				status = nfserr_inval;
1634 1635
				goto out;
			}
1636
			if (!creds_match) { /* case 9 */
1637
				status = nfserr_perm;
1638 1639 1640
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1641 1642 1643
				status = nfserr_not_same;
				goto out;
			}
1644 1645 1646 1647
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1648
		}
1649
		if (!creds_match) { /* case 3 */
1650 1651
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1652 1653 1654 1655 1656
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1657
		if (verfs_match) { /* case 2 */
1658
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1659 1660 1661 1662 1663
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1664 1665
	}

1666
	if (update) { /* case 7 */
1667 1668
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1669 1670
	}

1671
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1672
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1673 1674
		expire_client(unconf);

1675
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1676
out_new:
J
Jeff Layton 已提交
1677
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1678
	if (new == NULL) {
1679
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1680 1681
		goto out;
	}
1682
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1683

1684
	gen_clid(new, nn);
1685
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1686 1687 1688 1689
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1690
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1691 1692 1693
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1694
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1695 1696 1697 1698 1699
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1700 1701
}

J
J. Bruce Fields 已提交
1702
static __be32
1703
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1704
{
1705 1706
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1707 1708

	/* The slot is in use, and no response has been sent. */
1709 1710
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1711 1712 1713 1714
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1715
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1716
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1717
		return nfs_ok;
1718
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1719 1720 1721 1722
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1723 1724 1725 1726 1727 1728 1729
/*
 * 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
1730
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743
{
	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;
}

1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760
#define NFSD_MIN_REQ_HDR_SEQ_SZ	((\
			2 * 2 + /* credential,verifier: AUTH_NULL, length 0 */ \
			1 +	/* MIN tag is length with zero, only length */ \
			3 +	/* version, opcount, opcode */ \
			XDR_QUADLEN(NFS4_MAX_SESSIONID_LEN) + \
				/* seqid, slotID, slotID, cache */ \
			4 ) * sizeof(__be32))

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

J
J. Bruce Fields 已提交
1761
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1762 1763 1764 1765 1766
{
	return fchannel.maxreq_sz < NFSD_MIN_REQ_HDR_SEQ_SZ
		|| fchannel.maxresp_sz < NFSD_MIN_RESP_HDR_SEQ_SZ;
}

A
Andy Adamson 已提交
1767 1768 1769 1770 1771
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1772
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1773
	struct nfs4_client *conf, *unconf;
1774
	struct nfsd4_session *new;
1775
	struct nfsd4_conn *conn;
1776
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1777
	__be32 status = 0;
1778
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1779

1780 1781
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1782 1783
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1784 1785 1786 1787 1788 1789 1790
	new = alloc_session(&cr_ses->fore_channel);
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1791

A
Andy Adamson 已提交
1792
	nfs4_lock_state();
1793
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1794
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
A
Andy Adamson 已提交
1795 1796

	if (conf) {
1797 1798
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1799
		if (status == nfserr_replay_cache) {
1800
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1801
			goto out_free_conn;
1802
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1803
			status = nfserr_seq_misordered;
1804
			goto out_free_conn;
A
Andy Adamson 已提交
1805 1806
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1807
		struct nfs4_client *old;
A
Andy Adamson 已提交
1808
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1809
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1810
			status = nfserr_clid_inuse;
1811
			goto out_free_conn;
A
Andy Adamson 已提交
1812
		}
1813 1814
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1815 1816
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1817
			status = nfserr_seq_misordered;
1818
			goto out_free_conn;
A
Andy Adamson 已提交
1819
		}
1820
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
J
J. Bruce Fields 已提交
1821 1822 1823
		if (old)
			expire_client(old);
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1824 1825 1826
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1827
		goto out_free_conn;
A
Andy Adamson 已提交
1828
	}
1829
	status = nfs_ok;
1830 1831 1832 1833 1834 1835
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1836 1837 1838
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1839
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1840
	       NFS4_MAX_SESSIONID_LEN);
1841 1842
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1843
	cs_slot->sl_seqid++;
1844
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1845

1846 1847
	/* 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
1848 1849 1850 1851
out:
	nfs4_unlock_state();
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
1852 1853 1854 1855 1856
out_free_conn:
	free_conn(conn);
out_free_session:
	__free_session(new);
	goto out;
A
Andy Adamson 已提交
1857 1858
}

1859 1860 1861 1862 1863 1864 1865 1866
static bool nfsd4_last_compound_op(struct svc_rqst *rqstp)
{
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
	struct nfsd4_compoundargs *argp = rqstp->rq_argp;

	return argp->opcnt == resp->opcnt;
}

1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880
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;
}

1881 1882 1883
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1884
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1885

1886
	spin_lock(&nn->client_lock);
1887 1888
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1889
	spin_unlock(&nn->client_lock);
1890 1891 1892 1893 1894 1895

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1896 1897 1898 1899 1900
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1901
	struct nfsd4_conn *conn;
1902
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1903 1904 1905

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1906
	spin_lock(&nn->client_lock);
1907
	cstate->session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1908 1909 1910 1911 1912 1913
	/* Sorta weird: we only need the refcnt'ing because new_conn acquires
	 * client_lock iself: */
	if (cstate->session) {
		nfsd4_get_session(cstate->session);
		atomic_inc(&cstate->session->se_client->cl_refcount);
	}
1914
	spin_unlock(&nn->client_lock);
1915 1916 1917 1918
	if (!cstate->session)
		return nfserr_badsession;

	status = nfsd4_map_bcts_dir(&bcts->dir);
1919 1920 1921 1922 1923
	if (status)
		return status;
	conn = alloc_conn(rqstp, bcts->dir);
	if (!conn)
		return nfserr_jukebox;
1924
	nfsd4_init_conn(rqstp, conn, cstate->session);
1925
	return nfs_ok;
1926 1927
}

1928 1929 1930 1931 1932 1933 1934
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
1935 1936 1937 1938 1939
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1940
	struct nfsd4_session *ses;
J
J. Bruce Fields 已提交
1941
	__be32 status = nfserr_badsession;
1942
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1943 1944 1945 1946 1947 1948 1949 1950

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

1951
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1952 1953 1954
		if (!nfsd4_last_compound_op(r))
			return nfserr_not_only_op;
	}
B
Benny Halevy 已提交
1955
	dump_sessionid(__func__, &sessionid->sessionid);
1956
	spin_lock(&nn->client_lock);
1957
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
B
Benny Halevy 已提交
1958
	if (!ses) {
1959
		spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1960 1961 1962 1963
		goto out;
	}

	unhash_session(ses);
1964
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1965

1966
	nfs4_lock_state();
1967
	nfsd4_probe_callback_sync(ses->se_client);
1968
	nfs4_unlock_state();
1969

1970
	spin_lock(&nn->client_lock);
1971
	nfsd4_del_conns(ses);
1972
	nfsd4_put_session_locked(ses);
1973
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1974 1975 1976 1977
	status = nfs_ok;
out:
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
1978 1979
}

1980
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
1981 1982 1983 1984
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
1985
		if (c->cn_xprt == xpt) {
1986 1987 1988 1989 1990 1991
			return c;
		}
	}
	return NULL;
}

1992
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
1993 1994
{
	struct nfs4_client *clp = ses->se_client;
1995
	struct nfsd4_conn *c;
1996
	int ret;
1997 1998

	spin_lock(&clp->cl_lock);
1999
	c = __nfsd4_find_conn(new->cn_xprt, ses);
2000 2001 2002 2003 2004 2005 2006
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
2007 2008 2009 2010
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2011 2012 2013
	return;
}

2014 2015 2016 2017 2018 2019 2020
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
2021 2022 2023 2024 2025 2026 2027 2028
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
2029
__be32
B
Benny Halevy 已提交
2030
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2031 2032 2033
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2034
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2035 2036
	struct nfsd4_session *session;
	struct nfsd4_slot *slot;
2037
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2038
	__be32 status;
2039
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2040

2041 2042 2043
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2044 2045 2046 2047 2048 2049 2050 2051
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2052
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2053
	status = nfserr_badsession;
2054
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2055 2056 2057
	if (!session)
		goto out;

2058 2059 2060 2061
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
		goto out;

M
Mi Jinlong 已提交
2062 2063 2064 2065
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
		goto out;

B
Benny Halevy 已提交
2066
	status = nfserr_badslot;
2067
	if (seq->slotid >= session->se_fchannel.maxreqs)
B
Benny Halevy 已提交
2068 2069
		goto out;

2070
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2071 2072
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2073 2074 2075 2076 2077
	/* 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;

2078 2079
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2080
	if (status == nfserr_replay_cache) {
2081 2082 2083
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
			goto out;
B
Benny Halevy 已提交
2084 2085
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2086
		/* Return the cached reply status and set cstate->status
2087
		 * for nfsd4_proc_compound processing */
2088
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2089
		cstate->status = nfserr_replay_cache;
2090
		goto out;
B
Benny Halevy 已提交
2091 2092 2093 2094
	}
	if (status)
		goto out;

2095 2096
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2097

B
Benny Halevy 已提交
2098 2099
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2100
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2101 2102
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2103 2104
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2105 2106 2107 2108 2109

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

out:
2110
	/* Hold a session reference until done processing the compound. */
2111
	if (cstate->session) {
2112 2113
		struct nfs4_client *clp = session->se_client;

2114
		nfsd4_get_session(cstate->session);
2115
		atomic_inc(&clp->cl_refcount);
2116 2117
		switch (clp->cl_cb_state) {
		case NFSD4_CB_DOWN:
2118
			seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
2119 2120
			break;
		case NFSD4_CB_FAULT:
2121
			seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
2122
			break;
2123 2124
		default:
			seq->status_flags = 0;
2125
		}
2126
	}
2127
	kfree(conn);
2128
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2129 2130
	dprintk("%s: return %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
2131 2132
}

2133 2134 2135 2136
__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
2137
	__be32 status = 0;
2138
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2139 2140

	nfs4_lock_state();
2141
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2142
	conf = find_confirmed_client(&dc->clientid, true, nn);
2143 2144 2145 2146

	if (conf) {
		clp = conf;

2147
		if (!is_client_expired(conf) && client_has_state(conf)) {
2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170
			status = nfserr_clientid_busy;
			goto out;
		}

		/* rfc5661 18.50.3 */
		if (cstate->session && conf == cstate->session->se_client) {
			status = nfserr_clientid_busy;
			goto out;
		}
	} else if (unconf)
		clp = unconf;
	else {
		status = nfserr_stale_clientid;
		goto out;
	}

	expire_client(clp);
out:
	nfs4_unlock_state();
	dprintk("%s return %d\n", __func__, ntohl(status));
	return status;
}

2171 2172 2173
__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
2174
	__be32 status = 0;
2175

2176 2177 2178 2179 2180 2181 2182 2183 2184
	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;
	}
2185

2186
	nfs4_lock_state();
2187
	status = nfserr_complete_already;
2188 2189
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2190 2191 2192 2193
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2194 2195 2196 2197 2198 2199 2200
		/*
		 * 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.
		 */
2201 2202 2203
		goto out;

	status = nfs_ok;
2204
	nfsd4_client_record_create(cstate->session->se_client);
2205
out:
2206
	nfs4_unlock_state();
2207
	return status;
2208 2209
}

2210
__be32
2211 2212
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2213
{
2214
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2215
	nfs4_verifier		clverifier = setclid->se_verf;
2216
	struct nfs4_client	*conf, *unconf, *new;
2217
	__be32 			status;
2218 2219
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2220
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2221
	nfs4_lock_state();
2222
	conf = find_confirmed_client_by_name(&clname, nn);
2223
	if (conf) {
2224
		/* case 0: */
L
Linus Torvalds 已提交
2225
		status = nfserr_clid_inuse;
2226 2227
		if (clp_used_exchangeid(conf))
			goto out;
2228
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2229 2230 2231 2232 2233
			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
2234 2235 2236
			goto out;
		}
	}
2237
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2238 2239
	if (unconf)
		expire_client(unconf);
2240
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2241
	new = create_client(clname, rqstp, &clverifier);
2242 2243
	if (new == NULL)
		goto out;
2244
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2245
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2246
		copy_clid(new, conf);
2247
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2248
		gen_clid(new, nn);
2249
	new->cl_minorversion = 0;
2250
	gen_callback(new, setclid, rqstp);
2251
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
	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;
}


2262
__be32
2263 2264 2265
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2266
{
2267
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2268 2269
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2270
	__be32 status;
2271
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2272

2273
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2274 2275
		return nfserr_stale_clientid;
	nfs4_lock_state();
2276

2277
	conf = find_confirmed_client(clid, false, nn);
2278
	unconf = find_unconfirmed_client(clid, false, nn);
2279
	/*
2280 2281 2282 2283
	 * 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.
2284
	 */
2285 2286 2287 2288 2289
	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;
2290
	/* cases below refer to rfc 3530 section 14.2.34: */
2291 2292
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2293
			status = nfs_ok;
2294 2295 2296 2297 2298 2299
		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 */
2300 2301 2302
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2303
	} else { /* case 3: normal case; new or rebooted client */
2304
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2305
		if (conf)
2306 2307
			expire_client(conf);
		move_to_confirmed(unconf);
2308
		nfsd4_probe_callback(unconf);
2309
	}
L
Linus Torvalds 已提交
2310 2311 2312 2313 2314
out:
	nfs4_unlock_state();
	return status;
}

2315 2316 2317 2318 2319
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2320
/* OPEN Share state helper functions */
2321
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2322 2323 2324
{
	unsigned int hashval = file_hashval(ino);

2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
	atomic_set(&fp->fi_ref, 1);
	INIT_LIST_HEAD(&fp->fi_hash);
	INIT_LIST_HEAD(&fp->fi_stateids);
	INIT_LIST_HEAD(&fp->fi_delegations);
	fp->fi_inode = igrab(ino);
	fp->fi_had_conflict = false;
	fp->fi_lease = NULL;
	memset(fp->fi_fds, 0, sizeof(fp->fi_fds));
	memset(fp->fi_access, 0, sizeof(fp->fi_access));
	spin_lock(&recall_lock);
	list_add(&fp->fi_hash, &file_hashtbl[hashval]);
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2337 2338
}

N
NeilBrown 已提交
2339
static void
2340
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2341
{
N
NeilBrown 已提交
2342 2343
	if (*slab == NULL)
		return;
2344
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2345
	*slab = NULL;
L
Linus Torvalds 已提交
2346 2347
}

2348
void
L
Linus Torvalds 已提交
2349 2350
nfsd4_free_slabs(void)
{
2351 2352
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2353
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2354
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2355
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2356
}
L
Linus Torvalds 已提交
2357

2358
int
N
NeilBrown 已提交
2359 2360
nfsd4_init_slabs(void)
{
2361 2362 2363 2364 2365
	openowner_slab = kmem_cache_create("nfsd4_openowners",
			sizeof(struct nfs4_openowner), 0, 0, NULL);
	if (openowner_slab == NULL)
		goto out_nomem;
	lockowner_slab = kmem_cache_create("nfsd4_lockowners",
2366
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2367
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2368 2369
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2370
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2371 2372
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2373
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2374
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2375 2376
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2377
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2378
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2379 2380
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2381 2382 2383 2384 2385
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2386 2387
}

2388 2389 2390 2391 2392 2393 2394
void nfs4_free_openowner(struct nfs4_openowner *oo)
{
	kfree(oo->oo_owner.so_owner.data);
	kmem_cache_free(openowner_slab, oo);
}

void nfs4_free_lockowner(struct nfs4_lockowner *lo)
L
Linus Torvalds 已提交
2395
{
2396 2397
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2398 2399
}

2400
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2401
{
2402 2403 2404
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2405 2406
}

2407
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2408
{
L
Linus Torvalds 已提交
2409 2410
	struct nfs4_stateowner *sop;

2411
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2412 2413 2414 2415 2416
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2417
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2418
		return NULL;
2419 2420 2421
	}
	sop->so_owner.len = owner->len;

2422
	INIT_LIST_HEAD(&sop->so_stateids);
2423 2424 2425 2426 2427
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2428
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2429
{
2430 2431 2432
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2433
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2434 2435
}

2436
static struct nfs4_openowner *
2437
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2438
	struct nfs4_openowner *oo;
2439

2440 2441
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2442
		return NULL;
2443 2444
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2445
	oo->oo_flags = NFS4_OO_NEW;
2446
	oo->oo_time = 0;
2447
	oo->oo_last_closed_stid = NULL;
2448 2449 2450
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2451 2452
}

2453
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2454
	struct nfs4_openowner *oo = open->op_openowner;
2455
	struct nfs4_client *clp = oo->oo_owner.so_client;
L
Linus Torvalds 已提交
2456

2457
	init_stid(&stp->st_stid, clp, NFS4_OPEN_STID);
2458
	INIT_LIST_HEAD(&stp->st_lockowners);
2459
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2460
	list_add(&stp->st_perfile, &fp->fi_stateids);
2461
	stp->st_stateowner = &oo->oo_owner;
2462
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2463 2464 2465
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2466
	set_access(open->op_share_access, stp);
2467
	set_deny(open->op_share_deny, stp);
2468
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2469 2470
}

2471
static void
2472
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2473
{
2474 2475
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2476
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2477

2478
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2479
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2480 2481 2482
}

static int
2483 2484 2485 2486 2487 2488
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
2489 2490
}

2491
static struct nfs4_openowner *
2492 2493
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2494
{
2495 2496
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2497
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2498

2499
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2500 2501
		if (!so->so_is_open_owner)
			continue;
2502 2503
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2504 2505 2506
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2507 2508 2509
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520
	}
	return NULL;
}

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

2521
	spin_lock(&recall_lock);
L
Linus Torvalds 已提交
2522
	list_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2523 2524
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2525
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2526
			return fp;
2527
		}
L
Linus Torvalds 已提交
2528
	}
2529
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2530 2531 2532 2533 2534 2535 2536
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2537
static __be32
L
Linus Torvalds 已提交
2538 2539 2540 2541
nfs4_share_conflict(struct svc_fh *current_fh, unsigned int deny_type)
{
	struct inode *ino = current_fh->fh_dentry->d_inode;
	struct nfs4_file *fp;
2542
	struct nfs4_ol_stateid *stp;
2543
	__be32 ret;
L
Linus Torvalds 已提交
2544 2545 2546 2547

	dprintk("NFSD: nfs4_share_conflict\n");

	fp = find_file(ino);
2548 2549
	if (!fp)
		return nfs_ok;
2550
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2551
	/* Search for conflicting share reservations */
2552
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2553 2554
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2555
			goto out;
L
Linus Torvalds 已提交
2556
	}
2557 2558 2559 2560
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2561 2562
}

2563
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2564 2565 2566 2567 2568 2569 2570 2571 2572 2573
{
	/* We're assuming the state code never drops its reference
	 * without first removing the lease.  Since we're in this lease
	 * callback (and since the lease code is serialized by the kernel
	 * lock) we know the server hasn't removed the lease yet, we know
	 * it's safe to take a reference: */
	atomic_inc(&dp->dl_count);

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

2574
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2575 2576
	dp->dl_time = get_seconds();

2577 2578 2579
	nfsd4_cb_recall(dp);
}

2580
/* Called from break_lease() with lock_flocks() held. */
2581 2582
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2583 2584
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2585

2586 2587 2588 2589 2590 2591 2592 2593
	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;
	}
2594 2595
	/*
	 * We don't want the locks code to timeout the lease for us;
2596
	 * we'll remove it ourself if a delegation isn't returned
2597
	 * in time:
2598 2599
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2600

2601
	spin_lock(&recall_lock);
2602 2603 2604
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2605
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616
}

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

2617
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2618 2619
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2620 2621
};

2622 2623 2624 2625 2626 2627 2628 2629 2630 2631
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
2632

2633
__be32
A
Andy Adamson 已提交
2634
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2635
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2636 2637 2638 2639
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2640
	struct nfs4_openowner *oo = NULL;
2641
	__be32 status;
L
Linus Torvalds 已提交
2642

2643
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2644
		return nfserr_stale_clientid;
2645 2646 2647 2648 2649 2650 2651
	/*
	 * 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
2652

2653
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2654
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2655 2656
	open->op_openowner = oo;
	if (!oo) {
2657 2658
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2659
		if (clp == NULL)
2660
			return nfserr_expired;
2661
		goto new_owner;
L
Linus Torvalds 已提交
2662
	}
2663
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2664
		/* Replace unconfirmed owners without checking for replay. */
2665 2666 2667
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2668
		goto new_owner;
2669
	}
2670 2671 2672 2673 2674
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2675 2676 2677 2678 2679
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2680 2681 2682 2683
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2684
	return nfs_ok;
L
Linus Torvalds 已提交
2685 2686
}

2687
static inline __be32
N
NeilBrown 已提交
2688 2689 2690 2691 2692 2693 2694 2695
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;
}

2696
static int share_access_to_flags(u32 share_access)
2697
{
2698
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2699 2700
}

2701
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2702
{
2703
	struct nfs4_stid *ret;
2704

2705
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2706 2707 2708
	if (!ret)
		return NULL;
	return delegstateid(ret);
2709 2710
}

2711 2712 2713 2714 2715 2716
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;
}

2717
static __be32
2718
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2719 2720 2721
		struct nfs4_delegation **dp)
{
	int flags;
2722
	__be32 status = nfserr_bad_stateid;
2723

2724
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2725
	if (*dp == NULL)
2726
		goto out;
2727
	flags = share_access_to_flags(open->op_share_access);
2728 2729 2730
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2731
out:
2732
	if (!nfsd4_is_deleg_cur(open))
2733 2734 2735
		return nfs_ok;
	if (status)
		return status;
2736
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2737
	return nfs_ok;
2738 2739
}

2740
static __be32
2741
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2742
{
2743
	struct nfs4_ol_stateid *local;
2744
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2745

2746
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2747 2748 2749 2750
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2751
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2752 2753 2754
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2755
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2756
	}
2757
	return nfs_ok;
L
Linus Torvalds 已提交
2758 2759
}

2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770
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;
}

2771 2772
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2773 2774
{
	__be32 status;
2775 2776 2777
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788
	if (!fp->fi_fds[oflag]) {
		status = nfsd_open(rqstp, cur_fh, S_IFREG, access,
			&fp->fi_fds[oflag]);
		if (status)
			return status;
	}
	nfs4_file_get_access(fp, oflag);

	return nfs_ok;
}

2789
static inline __be32
L
Linus Torvalds 已提交
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799
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))
2800
		return nfserr_inval;
L
Linus Torvalds 已提交
2801 2802 2803
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2804
static __be32
2805
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
2806
{
2807
	u32 op_share_access = open->op_share_access;
2808
	bool new_access;
2809
	__be32 status;
L
Linus Torvalds 已提交
2810

2811
	new_access = !test_access(op_share_access, stp);
2812
	if (new_access) {
2813
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2814 2815
		if (status)
			return status;
2816
	}
L
Linus Torvalds 已提交
2817 2818
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2819
		if (new_access) {
2820
			int oflag = nfs4_access_to_omode(op_share_access);
2821 2822
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2823 2824 2825
		return status;
	}
	/* remember the open */
2826
	set_access(op_share_access, stp);
2827
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2828 2829 2830 2831 2832 2833

	return nfs_ok;
}


static void
2834
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2835
{
2836
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2837 2838
}

2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
/* 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;
}

2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863
static struct file_lock *nfs4_alloc_init_lease(struct nfs4_delegation *dp, int flag)
{
	struct file_lock *fl;

	fl = locks_alloc_lock();
	if (!fl)
		return NULL;
	locks_init_lock(fl);
	fl->fl_lmops = &nfsd_lease_mng_ops;
	fl->fl_flags = FL_LEASE;
	fl->fl_type = flag == NFS4_OPEN_DELEGATE_READ? F_RDLCK: F_WRLCK;
	fl->fl_end = OFFSET_MAX;
2864
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2865 2866 2867 2868
	fl->fl_pid = current->tgid;
	return fl;
}

2869 2870
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2871
	struct nfs4_file *fp = dp->dl_file;
2872 2873 2874 2875 2876 2877
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2878
	fl->fl_file = find_readable_file(fp);
2879
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2880
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2881
	if (status) {
2882
		list_del_init(&dp->dl_perclnt);
2883 2884 2885
		locks_free_lock(fl);
		return -ENOMEM;
	}
2886
	fp->fi_lease = fl;
A
Al Viro 已提交
2887
	fp->fi_deleg_file = get_file(fl->fl_file);
2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
	atomic_set(&fp->fi_delegees, 1);
	list_add(&dp->dl_perfile, &fp->fi_delegations);
	return 0;
}

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

	if (!fp->fi_lease)
		return nfs4_setlease(dp, flag);
	spin_lock(&recall_lock);
	if (fp->fi_had_conflict) {
		spin_unlock(&recall_lock);
		return -EAGAIN;
	}
	atomic_inc(&fp->fi_delegees);
	list_add(&dp->dl_perfile, &fp->fi_delegations);
	spin_unlock(&recall_lock);
2907
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2908 2909 2910
	return 0;
}

2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
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:
2927
			WARN_ON_ONCE(1);
2928 2929 2930 2931
		}
	}
}

L
Linus Torvalds 已提交
2932 2933 2934 2935
/*
 * Attempt to hand out a delegation.
 */
static void
2936 2937
nfs4_open_delegation(struct net *net, struct svc_fh *fh,
		     struct nfsd4_open *open, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
2938 2939
{
	struct nfs4_delegation *dp;
2940
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2941
	int cb_up;
2942
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2943

2944
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2945
	flag = NFS4_OPEN_DELEGATE_NONE;
2946 2947 2948
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2949
			if (!cb_up)
2950 2951 2952 2953 2954 2955 2956 2957
				open->op_recall = 1;
			flag = open->op_delegate_type;
			if (flag == NFS4_OPEN_DELEGATE_NONE)
				goto out;
			break;
		case NFS4_OPEN_CLAIM_NULL:
			/* Let's not give out any delegations till everyone's
			 * had the chance to reclaim theirs.... */
2958
			if (locks_in_grace(net))
2959
				goto out;
2960
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2961 2962 2963 2964 2965 2966 2967 2968 2969
				goto out;
			if (open->op_share_access & NFS4_SHARE_ACCESS_WRITE)
				flag = NFS4_OPEN_DELEGATE_WRITE;
			else
				flag = NFS4_OPEN_DELEGATE_READ;
			break;
		default:
			goto out;
	}
L
Linus Torvalds 已提交
2970

2971
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2972 2973
	if (dp == NULL)
		goto out_no_deleg;
2974
	status = nfs4_set_delegation(dp, flag);
2975
	if (status)
2976
		goto out_free;
L
Linus Torvalds 已提交
2977

2978
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2979

2980
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
2981
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2982 2983
out:
	open->op_delegate_type = flag;
2984 2985 2986 2987 2988
	if (flag == NFS4_OPEN_DELEGATE_NONE) {
		if (open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS &&
		    open->op_delegate_type != NFS4_OPEN_DELEGATE_NONE)
			dprintk("NFSD: WARNING: refusing delegation reclaim\n");

2989 2990 2991
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
2992
	}
2993 2994
	return;
out_free:
2995
	nfs4_put_delegation(dp);
2996 2997 2998
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
2999 3000
}

3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018
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
3019 3020 3021
/*
 * called with nfs4_lock_state() held.
 */
3022
__be32
L
Linus Torvalds 已提交
3023 3024
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3025
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3026
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3027 3028
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3029
	struct nfs4_ol_stateid *stp = NULL;
3030
	struct nfs4_delegation *dp = NULL;
3031
	__be32 status;
L
Linus Torvalds 已提交
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041

	/*
	 * Lookup file; if found, lookup stateid and check open request,
	 * and check for delegations in the process of being recalled.
	 * If not found, create the nfs4_file struct
	 */
	fp = find_file(ino);
	if (fp) {
		if ((status = nfs4_check_open(fp, open, &stp)))
			goto out;
3042
		status = nfs4_check_deleg(cl, fp, open, &dp);
3043 3044
		if (status)
			goto out;
L
Linus Torvalds 已提交
3045
	} else {
3046
		status = nfserr_bad_stateid;
3047
		if (nfsd4_is_deleg_cur(open))
3048
			goto out;
3049
		status = nfserr_jukebox;
3050 3051 3052
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3053 3054 3055 3056 3057 3058 3059 3060
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3061
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3062 3063 3064
		if (status)
			goto out;
	} else {
3065
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3066 3067 3068
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3069
		if (status)
L
Linus Torvalds 已提交
3070
			goto out;
3071 3072
		stp = open->op_stp;
		open->op_stp = NULL;
3073
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3074
	}
3075 3076
	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
3077

3078
	if (nfsd4_has_session(&resp->cstate)) {
3079
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3080

3081 3082 3083 3084 3085 3086 3087
		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
3088 3089 3090 3091
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3092
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3093
nodeleg:
L
Linus Torvalds 已提交
3094 3095
	status = nfs_ok;

3096
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3097
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3098
out:
3099 3100
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3101 3102
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3103

3104 3105
	if (fp)
		put_nfs4_file(fp);
3106
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3107
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3108 3109 3110 3111
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3112
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3113
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3114 3115 3116 3117 3118
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133
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;
		}
	}
3134 3135
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3136
	if (open->op_stp)
3137
		free_generic_stateid(open->op_stp);
3138 3139
}

3140
__be32
3141 3142
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3143 3144
{
	struct nfs4_client *clp;
3145
	__be32 status;
3146
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3147 3148 3149 3150 3151

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
	status = nfserr_stale_clientid;
3152
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
3153
		goto out;
3154
	clp = find_confirmed_client(clid, cstate->minorversion, nn);
L
Linus Torvalds 已提交
3155 3156 3157 3158 3159 3160 3161
	status = nfserr_expired;
	if (clp == NULL) {
		/* We assume the client took too long to RENEW. */
		dprintk("nfsd4_renew: clientid not found!\n");
		goto out;
	}
	status = nfserr_cb_path_down;
3162
	if (!list_empty(&clp->cl_delegations)
3163
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3164 3165 3166 3167 3168 3169 3170
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3171
static void
3172
nfsd4_end_grace(struct nfsd_net *nn)
3173
{
3174
	/* do nothing if grace period already ended */
3175
	if (nn->grace_ended)
3176 3177
		return;

3178
	dprintk("NFSD: end of grace period\n");
3179
	nn->grace_ended = true;
3180
	nfsd4_record_grace_done(nn, nn->boot_time);
3181
	locks_end_grace(&nn->nfsd4_manager);
3182 3183 3184 3185 3186 3187
	/*
	 * 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:
	 */
	nfsd4_grace = nfsd4_lease;
3188 3189
}

3190
static time_t
3191
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3192 3193
{
	struct nfs4_client *clp;
3194
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3195 3196
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3197 3198 3199
	time_t cutoff = get_seconds() - nfsd4_lease;
	time_t t, clientid_val = nfsd4_lease;
	time_t u, test_val = nfsd4_lease;
L
Linus Torvalds 已提交
3200 3201 3202 3203

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3204
	nfsd4_end_grace(nn);
3205
	INIT_LIST_HEAD(&reaplist);
3206
	spin_lock(&nn->client_lock);
3207
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3208 3209 3210 3211 3212 3213 3214
		clp = list_entry(pos, struct nfs4_client, cl_lru);
		if (time_after((unsigned long)clp->cl_time, (unsigned long)cutoff)) {
			t = clp->cl_time - cutoff;
			if (clientid_val > t)
				clientid_val = t;
			break;
		}
3215 3216 3217 3218 3219 3220 3221
		if (atomic_read(&clp->cl_refcount)) {
			dprintk("NFSD: client in use (clientid %08x)\n",
				clp->cl_clientid.cl_id);
			continue;
		}
		unhash_client_locked(clp);
		list_add(&clp->cl_lru, &reaplist);
3222
	}
3223
	spin_unlock(&nn->client_lock);
3224 3225
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3226 3227 3228 3229 3230 3231 3232
		dprintk("NFSD: purging unused client (clientid %08x)\n",
			clp->cl_clientid.cl_id);
		expire_client(clp);
	}
	spin_lock(&recall_lock);
	list_for_each_safe(pos, next, &del_recall_lru) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
3233 3234
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247
		if (time_after((unsigned long)dp->dl_time, (unsigned long)cutoff)) {
			u = dp->dl_time - cutoff;
			if (test_val > u)
				test_val = u;
			break;
		}
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	list_for_each_safe(pos, next, &reaplist) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}
3248
	test_val = nfsd4_lease;
3249
	list_for_each_safe(pos, next, &nn->close_lru) {
3250 3251 3252
		oo = container_of(pos, struct nfs4_openowner, oo_close_lru);
		if (time_after((unsigned long)oo->oo_time, (unsigned long)cutoff)) {
			u = oo->oo_time - cutoff;
L
Linus Torvalds 已提交
3253 3254 3255 3256
			if (test_val > u)
				test_val = u;
			break;
		}
3257
		release_openowner(oo);
L
Linus Torvalds 已提交
3258 3259 3260 3261 3262 3263 3264
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3265 3266 3267 3268
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3269
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3270 3271
{
	time_t t;
3272 3273 3274 3275
	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
3276

3277
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3278
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3279
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3280 3281
}

3282
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3283
{
3284 3285 3286
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3287 3288 3289
}

static int
3290
STALE_STATEID(stateid_t *stateid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3291
{
3292
	if (stateid->si_opaque.so_clid.cl_boot == nn->boot_time)
3293 3294
		return 0;
	dprintk("NFSD: stale stateid " STATEID_FMT "!\n",
3295
		STATEID_VAL(stateid));
3296
	return 1;
L
Linus Torvalds 已提交
3297 3298 3299
}

static inline int
3300
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3301
{
3302 3303 3304
	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
3305 3306 3307
}

static inline int
3308
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3309
{
3310 3311
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3312 3313 3314
}

static
3315
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3316
{
3317
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3318

3319 3320 3321
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3322
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3323
                goto out;
3324
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3325 3326 3327 3328 3329 3330
                goto out;
	status = nfs_ok;
out:
	return status;
}

3331
static inline __be32
3332
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3333
{
3334
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3335
		return nfs_ok;
3336
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3337
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352
		 * 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
3353
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3354
{
3355
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3356 3357
}

3358 3359 3360 3361 3362 3363
/* Returns true iff a is later than b: */
static bool stateid_generation_after(stateid_t *a, stateid_t *b)
{
	return (s32)a->si_generation - (s32)b->si_generation > 0;
}

J
J. Bruce Fields 已提交
3364
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3365
{
A
Andy Adamson 已提交
3366 3367 3368 3369
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3370
	if (has_session && in->si_generation == 0)
3371 3372 3373 3374
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3375

3376
	/* If the client sends us a stateid from the future, it's buggy: */
3377
	if (stateid_generation_after(in, ref))
3378 3379
		return nfserr_bad_stateid;
	/*
3380 3381 3382 3383 3384 3385 3386 3387
	 * 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:
3388
	 */
3389
	return nfserr_old_stateid;
3390 3391
}

3392
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3393
{
3394 3395 3396
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3397

3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408
	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
	/* Client debugging aid. */
	if (!same_clid(&stateid->si_opaque.so_clid, &cl->cl_clientid)) {
		char addr_str[INET6_ADDRSTRLEN];
		rpc_ntop((struct sockaddr *)&cl->cl_addr, addr_str,
				 sizeof(addr_str));
		pr_warn_ratelimited("NFSD: client %s testing state ID "
					"with incorrect client ID\n", addr_str);
		return nfserr_bad_stateid;
	}
3409
	s = find_stateid(cl, stateid);
3410
	if (!s)
3411
		return nfserr_bad_stateid;
3412
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3413
	if (status)
3414 3415 3416 3417 3418
		return status;
	if (!(s->sc_type & (NFS4_OPEN_STID | NFS4_LOCK_STID)))
		return nfs_ok;
	ols = openlockstateid(s);
	if (ols->st_stateowner->so_is_open_owner
3419
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3420 3421
		return nfserr_bad_stateid;
	return nfs_ok;
3422 3423
}

3424 3425 3426
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3427 3428 3429 3430 3431
{
	struct nfs4_client *cl;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
3432
	if (STALE_STATEID(stateid, nn))
3433
		return nfserr_stale_stateid;
3434
	cl = find_confirmed_client(&stateid->si_opaque.so_clid, sessions, nn);
3435 3436 3437 3438 3439 3440 3441 3442 3443
	if (!cl)
		return nfserr_expired;
	*s = find_stateid_by_type(cl, stateid, typemask);
	if (!*s)
		return nfserr_bad_stateid;
	return nfs_ok;

}

L
Linus Torvalds 已提交
3444 3445 3446
/*
* Checks for stateid operations
*/
3447
__be32
3448
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3449
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3450
{
3451
	struct nfs4_stid *s;
3452
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3453
	struct nfs4_delegation *dp = NULL;
3454
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3455
	struct inode *ino = current_fh->fh_dentry->d_inode;
3456
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3457
	__be32 status;
L
Linus Torvalds 已提交
3458 3459 3460 3461

	if (filpp)
		*filpp = NULL;

3462
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3463 3464 3465
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3466
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3467

3468 3469
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3470 3471
	if (status)
		return status;
3472 3473 3474
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3475 3476
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3477
		dp = delegstateid(s);
3478 3479 3480
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3481
		if (filpp) {
3482
			*filpp = dp->dl_file->fi_deleg_file;
3483 3484 3485 3486 3487
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3488
		}
3489 3490 3491
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3492
		stp = openlockstateid(s);
3493 3494
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3495
			goto out;
3496
		if (stp->st_stateowner->so_is_open_owner
3497
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3498
			goto out;
3499 3500
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3501
			goto out;
3502 3503 3504 3505 3506 3507
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3508 3509 3510
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3511 3512 3513 3514 3515 3516
	}
	status = nfs_ok;
out:
	return status;
}

3517
static __be32
3518
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3519
{
3520
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3521 3522 3523 3524 3525
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3526 3527 3528 3529 3530 3531 3532
/*
 * 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)
{
3533 3534 3535 3536 3537
	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)
3538 3539
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3540 3541
	nfs4_unlock_state();

3542 3543 3544
	return nfs_ok;
}

3545 3546 3547 3548 3549
__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
3550
	struct nfs4_stid *s;
3551
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3552
	__be32 ret = nfserr_bad_stateid;
3553 3554

	nfs4_lock_state();
3555
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3556
	if (!s)
3557
		goto out;
J
J. Bruce Fields 已提交
3558 3559
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3560 3561
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3562 3563 3564 3565 3566 3567 3568 3569 3570
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
		ret = check_stateid_generation(stateid, &s->sc_stateid, 1);
		if (ret)
			goto out;
		if (s->sc_type == NFS4_LOCK_STID)
			ret = nfsd4_free_lock_stateid(openlockstateid(s));
		else
			ret = nfserr_locks_held;
3571 3572 3573
		break;
	default:
		ret = nfserr_bad_stateid;
3574 3575 3576 3577 3578 3579
	}
out:
	nfs4_unlock_state();
	return ret;
}

3580 3581 3582 3583 3584 3585
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
3586

3587
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3588 3589 3590 3591 3592 3593 3594 3595
{
	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;
3596 3597 3598 3599 3600 3601 3602 3603 3604 3605
	if (stp->st_stid.sc_type == NFS4_CLOSED_STID)
		/*
		 * "Closed" stateid's exist *only* to return
		 * nfserr_replay_me from the previous step.
		 */
		return nfserr_bad_stateid;
	status = check_stateid_generation(stateid, &stp->st_stid.sc_stateid, nfsd4_has_session(cstate));
	if (status)
		return status;
	return nfs4_check_fh(current_fh, stp);
3606 3607
}

L
Linus Torvalds 已提交
3608 3609 3610
/* 
 * Checks for sequence id mutating operations. 
 */
3611
static __be32
3612
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3613
			 stateid_t *stateid, char typemask,
3614 3615
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3616
{
3617
	__be32 status;
3618
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3619

3620 3621
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3622

L
Linus Torvalds 已提交
3623
	*stpp = NULL;
3624 3625
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3626 3627
	if (status)
		return status;
3628
	*stpp = openlockstateid(s);
3629
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3630

3631 3632
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3633

3634 3635
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)
3636 3637 3638
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3639

3640
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3641
						NFS4_OPEN_STID, stpp, nn);
3642 3643
	if (status)
		return status;
3644
	oo = openowner((*stpp)->st_stateowner);
3645
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3646 3647
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3648 3649
}

3650
__be32
3651
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3652
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3653
{
3654
	__be32 status;
3655
	struct nfs4_openowner *oo;
3656
	struct nfs4_ol_stateid *stp;
3657
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3658 3659

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3660 3661
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3662

3663
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3664 3665
	if (status)
		return status;
L
Linus Torvalds 已提交
3666 3667 3668

	nfs4_lock_state();

3669
	status = nfs4_preprocess_seqid_op(cstate,
3670
					oc->oc_seqid, &oc->oc_req_stateid,
3671
					NFS4_OPEN_STID, &stp, nn);
3672
	if (status)
3673
		goto out;
3674
	oo = openowner(stp->st_stateowner);
3675
	status = nfserr_bad_stateid;
3676
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3677
		goto out;
3678
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3679 3680
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3681
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3682
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3683

3684
	nfsd4_client_record_create(oo->oo_owner.so_client);
3685
	status = nfs_ok;
L
Linus Torvalds 已提交
3686
out:
3687 3688
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3689 3690 3691
	return status;
}

J
J. Bruce Fields 已提交
3692
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3693
{
3694
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3695 3696
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3697
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3698
}
3699

J
J. Bruce Fields 已提交
3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713
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:
3714
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3715 3716 3717 3718
	}
}

static void
3719
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3720 3721 3722 3723
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3724
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3725 3726 3727
	}
}

3728
__be32
3729 3730
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3731
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3732
{
3733
	__be32 status;
3734
	struct nfs4_ol_stateid *stp;
3735
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3736 3737

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3738 3739
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3740

3741
	/* We don't yet support WANT bits: */
3742 3743 3744
	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
3745 3746

	nfs4_lock_state();
3747
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3748
					&od->od_stateid, &stp, nn);
3749
	if (status)
L
Linus Torvalds 已提交
3750 3751
		goto out; 
	status = nfserr_inval;
3752 3753
	if (!test_access(od->od_share_access, stp)) {
		dprintk("NFSD: access not a subset current bitmap: 0x%lx, input access=%08x\n",
L
Linus Torvalds 已提交
3754 3755 3756
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3757
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3758 3759 3760 3761
		dprintk("NFSD:deny not a subset current bitmap: 0x%lx, input deny=%08x\n",
			stp->st_deny_bmap, od->od_share_deny);
		goto out;
	}
J
J. Bruce Fields 已提交
3762
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3763

3764
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3765

3766 3767
	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
3768 3769
	status = nfs_ok;
out:
3770 3771
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3772 3773 3774
	return status;
}

3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791
void nfsd4_purge_closed_stateid(struct nfs4_stateowner *so)
{
	struct nfs4_openowner *oo;
	struct nfs4_ol_stateid *s;

	if (!so->so_is_open_owner)
		return;
	oo = openowner(so);
	s = oo->oo_last_closed_stid;
	if (!s)
		return;
	if (!(oo->oo_flags & NFS4_OO_PURGE_CLOSE)) {
		/* Release the last_closed_stid on the next seqid bump: */
		oo->oo_flags |= NFS4_OO_PURGE_CLOSE;
		return;
	}
	oo->oo_flags &= ~NFS4_OO_PURGE_CLOSE;
3792 3793 3794 3795 3796 3797 3798
	release_last_closed_stateid(oo);
}

static void nfsd4_close_open_stateid(struct nfs4_ol_stateid *s)
{
	unhash_open_stateid(s);
	s->st_stid.sc_type = NFS4_CLOSED_STID;
3799 3800
}

L
Linus Torvalds 已提交
3801 3802 3803
/*
 * nfs4_unlock_state() called after encode
 */
3804
__be32
3805
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3806
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3807
{
3808
	__be32 status;
3809
	struct nfs4_openowner *oo;
3810
	struct nfs4_ol_stateid *stp;
3811 3812
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3813 3814

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3815 3816
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3817 3818

	nfs4_lock_state();
3819 3820 3821
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3822
					&stp, nn);
3823
	if (status)
L
Linus Torvalds 已提交
3824
		goto out; 
3825
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3826
	status = nfs_ok;
3827 3828
	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
3829

3830
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3831
	release_last_closed_stateid(oo);
3832
	oo->oo_last_closed_stid = stp;
3833

3834 3835 3836 3837 3838 3839 3840 3841 3842 3843
	if (list_empty(&oo->oo_owner.so_stateids)) {
		if (cstate->minorversion) {
			release_openowner(oo);
			cstate->replay_owner = NULL;
		} else {
			/*
			 * In the 4.0 case we need to keep the owners around a
			 * little while to handle CLOSE replay.
			 */
			if (list_empty(&oo->oo_owner.so_stateids))
3844
				move_to_close_lru(oo, SVC_NET(rqstp));
3845 3846
		}
	}
L
Linus Torvalds 已提交
3847
out:
3848 3849
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3850 3851 3852
	return status;
}

3853
__be32
3854 3855
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3856
{
3857 3858
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3859
	struct nfs4_stid *s;
3860
	__be32 status;
3861
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3862

3863
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3864
		return status;
L
Linus Torvalds 已提交
3865 3866

	nfs4_lock_state();
3867 3868
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3869
	if (status)
3870
		goto out;
3871
	dp = delegstateid(s);
3872
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3873 3874 3875 3876
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3877
out:
3878 3879
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3880 3881 3882 3883 3884 3885
	return status;
}


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

3886
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3887

B
Benny Halevy 已提交
3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902
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;

3903
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3904 3905 3906 3907
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3908
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3909 3910 3911
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3912
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931
}

/*
 * 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;
}

3932 3933
/* 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. */
3934
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3935
};
L
Linus Torvalds 已提交
3936 3937 3938 3939

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3940
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3941

3942
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3943 3944 3945
		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);
3946 3947 3948
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3949 3950
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3951
	} else {
3952 3953 3954
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3955 3956
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3957 3958
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3959 3960
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3961 3962 3963 3964 3965 3966
		deny->ld_length = fl->fl_end - fl->fl_start + 1;        
	deny->ld_type = NFS4_READ_LT;
	if (fl->fl_type != F_RDLCK)
		deny->ld_type = NFS4_WRITE_LT;
}

J
J. Bruce Fields 已提交
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
static bool same_lockowner_ino(struct nfs4_lockowner *lo, struct inode *inode, clientid_t *clid, struct xdr_netobj *owner)
{
	struct nfs4_ol_stateid *lst;

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

3978 3979
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3980
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3981
{
3982
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
3983
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3984

3985
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
3986 3987
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
3988 3989 3990 3991
	}
	return NULL;
}

3992
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
3993
{
3994 3995 3996
	struct inode *inode = open_stp->st_file->fi_inode;
	unsigned int inohash = lockowner_ino_hashval(inode,
			clp->cl_clientid.cl_id, &lo->lo_owner.so_owner);
3997
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
3998

3999
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
4000
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
4001
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
4002 4003
}

L
Linus Torvalds 已提交
4004 4005 4006
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4007
 * occurred. 
L
Linus Torvalds 已提交
4008
 *
4009
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4010 4011
 */

4012
static struct nfs4_lockowner *
4013
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4014
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4015

4016 4017
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4018
		return NULL;
4019 4020
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4021 4022
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4023 4024 4025
	lo->lo_owner.so_seqid = lock->lk_new_lock_seqid + 1;
	hash_lockowner(lo, strhashval, clp, open_stp);
	return lo;
L
Linus Torvalds 已提交
4026 4027
}

4028 4029
static struct nfs4_ol_stateid *
alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct nfs4_ol_stateid *open_stp)
L
Linus Torvalds 已提交
4030
{
4031
	struct nfs4_ol_stateid *stp;
4032
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4033

4034
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4035
	if (stp == NULL)
J
J. Bruce Fields 已提交
4036
		return NULL;
4037
	init_stid(&stp->st_stid, clp, NFS4_LOCK_STID);
4038
	list_add(&stp->st_perfile, &fp->fi_stateids);
4039 4040
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4041
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4042
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4043
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4044
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4045
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4046 4047 4048
	return stp;
}

4049
static int
L
Linus Torvalds 已提交
4050 4051
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4052
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4053 4054 4055
	     LOFF_OVERFLOW(offset, length)));
}

4056
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4057 4058 4059 4060
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4061
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4062 4063
		return;
	nfs4_file_get_access(fp, oflag);
4064
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4065 4066
}

J
J. Bruce Fields 已提交
4067
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)
4068 4069 4070 4071 4072 4073
{
	struct nfs4_file *fi = ost->st_file;
	struct nfs4_openowner *oo = openowner(ost->st_stateowner);
	struct nfs4_client *cl = oo->oo_owner.so_client;
	struct nfs4_lockowner *lo;
	unsigned int strhashval;
4074
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4075

4076 4077
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4078 4079 4080 4081 4082 4083 4084 4085
	if (lo) {
		if (!cstate->minorversion)
			return nfserr_bad_seqid;
		/* XXX: a lockowner always has exactly one stateid: */
		*lst = list_first_entry(&lo->lo_owner.so_stateids,
				struct nfs4_ol_stateid, st_perstateowner);
		return nfs_ok;
	}
4086
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099
			&lock->v.new.owner);
	lo = alloc_init_lock_stateowner(strhashval, cl, ost, lock);
	if (lo == NULL)
		return nfserr_jukebox;
	*lst = alloc_init_lock_stateid(lo, fi, ost);
	if (*lst == NULL) {
		release_lockowner(lo);
		return nfserr_jukebox;
	}
	*new = true;
	return nfs_ok;
}

L
Linus Torvalds 已提交
4100 4101 4102
/*
 *  LOCK operation 
 */
4103
__be32
4104
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4105
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4106
{
4107 4108
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4109
	struct nfs4_ol_stateid *lock_stp;
4110
	struct file *filp = NULL;
4111 4112
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4113
	__be32 status = 0;
4114
	bool new_state = false;
4115
	int lkflg;
4116
	int err;
4117 4118
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4119 4120 4121 4122 4123 4124 4125 4126

	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;

4127
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4128
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4129 4130 4131 4132
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4133 4134 4135
	nfs4_lock_state();

	if (lock->lk_is_new) {
4136
		struct nfs4_ol_stateid *open_stp = NULL;
4137 4138 4139 4140 4141 4142 4143

		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
4144
		status = nfserr_stale_clientid;
4145
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4146 4147 4148
			goto out;

		/* validate and update open stateid and open seqid */
4149
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4150 4151
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4152
					&open_stp, nn);
4153
		if (status)
L
Linus Torvalds 已提交
4154
			goto out;
4155
		open_sop = openowner(open_stp->st_stateowner);
4156
		status = nfserr_bad_stateid;
4157
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4158 4159
						&lock->v.new.clientid))
			goto out;
4160 4161
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4162
	} else
4163
		status = nfs4_preprocess_seqid_op(cstate,
4164 4165
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4166
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4167 4168
	if (status)
		goto out;
4169
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4170

4171 4172 4173 4174 4175
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4176
	status = nfserr_grace;
4177
	if (locks_in_grace(net) && !lock->lk_reclaim)
4178 4179
		goto out;
	status = nfserr_no_grace;
4180
	if (!locks_in_grace(net) && lock->lk_reclaim)
4181 4182
		goto out;

4183 4184 4185 4186 4187 4188 4189 4190
	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
4191 4192 4193
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4194 4195 4196
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4197
			file_lock->fl_type = F_RDLCK;
4198
			break;
L
Linus Torvalds 已提交
4199 4200
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4201 4202 4203
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4204
			file_lock->fl_type = F_WRLCK;
4205
			break;
L
Linus Torvalds 已提交
4206 4207 4208 4209
		default:
			status = nfserr_inval;
		goto out;
	}
4210 4211 4212 4213
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228
	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
4229

4230
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4231
	switch (-err) {
L
Linus Torvalds 已提交
4232
	case 0: /* success! */
4233 4234
		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
4235
				sizeof(stateid_t));
4236
		status = 0;
4237 4238 4239 4240
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4241
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4242
		break;
L
Linus Torvalds 已提交
4243 4244
	case (EDEADLK):
		status = nfserr_deadlock;
4245
		break;
4246
	default:
4247
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4248
		status = nfserrno(err);
4249
		break;
L
Linus Torvalds 已提交
4250 4251
	}
out:
4252
	if (status && new_state)
4253
		release_lockowner(lock_sop);
4254 4255
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4256 4257 4258 4259
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4260 4261 4262
	return status;
}

4263 4264 4265 4266 4267 4268
/*
 * 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.)
 */
4269
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4270 4271
{
	struct file *file;
4272 4273 4274 4275 4276
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4277 4278 4279
	return err;
}

L
Linus Torvalds 已提交
4280 4281 4282
/*
 * LOCKT operation
 */
4283
__be32
4284 4285
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4286 4287
{
	struct inode *inode;
4288
	struct file_lock *file_lock = NULL;
4289
	struct nfs4_lockowner *lo;
4290
	__be32 status;
4291
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4292

4293
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4294 4295 4296 4297 4298 4299 4300 4301
		return nfserr_grace;

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

	nfs4_lock_state();

	status = nfserr_stale_clientid;
4302
	if (!nfsd4_has_session(cstate) && STALE_CLIENTID(&lockt->lt_clientid, nn))
L
Linus Torvalds 已提交
4303 4304
		goto out;

4305
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4306 4307
		goto out;

4308
	inode = cstate->current_fh.fh_dentry->d_inode;
4309 4310 4311 4312 4313 4314 4315
	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
4316 4317 4318
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4319
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4320 4321 4322
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4323
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4324 4325
		break;
		default:
4326
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4327 4328 4329 4330
			status = nfserr_inval;
		goto out;
	}

4331
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4332
	if (lo)
4333 4334 4335
		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
4336

4337 4338
	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
4339

4340
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4341

4342
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4343
	if (status)
4344
		goto out;
4345

4346
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4347
		status = nfserr_denied;
4348
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4349 4350 4351
	}
out:
	nfs4_unlock_state();
4352 4353
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4354 4355 4356
	return status;
}

4357
__be32
4358
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4359
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4360
{
4361
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4362
	struct file *filp = NULL;
4363
	struct file_lock *file_lock = NULL;
4364
	__be32 status;
4365
	int err;
4366 4367
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4368 4369 4370 4371 4372 4373 4374 4375 4376
	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();
									        
4377
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4378 4379
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4380
	if (status)
L
Linus Torvalds 已提交
4381
		goto out;
4382 4383 4384 4385 4386
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404
	file_lock = locks_alloc_lock();
	if (!file_lock) {
		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
		status = nfserr_jukebox;
		goto out;
	}
	locks_init_lock(file_lock);
	file_lock->fl_type = F_UNLCK;
	file_lock->fl_owner = (fl_owner_t)lockowner(stp->st_stateowner);
	file_lock->fl_pid = current->tgid;
	file_lock->fl_file = filp;
	file_lock->fl_flags = FL_POSIX;
	file_lock->fl_lmops = &nfsd_posix_mng_ops;
	file_lock->fl_start = locku->lu_offset;

	file_lock->fl_end = last_byte_offset(locku->lu_offset,
						locku->lu_length);
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4405 4406 4407 4408

	/*
	*  Try to unlock the file in the VFS.
	*/
4409
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4410
	if (err) {
4411
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4412 4413 4414 4415 4416
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4417 4418
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&locku->lu_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
L
Linus Torvalds 已提交
4419 4420

out:
4421 4422
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4423 4424
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4425 4426 4427
	return status;

out_nfserr:
4428
	status = nfserrno(err);
L
Linus Torvalds 已提交
4429 4430 4431 4432 4433 4434 4435 4436 4437
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4438
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4439 4440
{
	struct file_lock **flpp;
4441
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4442 4443
	int status = 0;

4444
	lock_flocks();
L
Linus Torvalds 已提交
4445
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4446
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4447 4448
			status = 1;
			goto out;
4449
		}
L
Linus Torvalds 已提交
4450 4451
	}
out:
4452
	unlock_flocks();
L
Linus Torvalds 已提交
4453 4454 4455
	return status;
}

4456
__be32
4457 4458 4459
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4460 4461
{
	clientid_t *clid = &rlockowner->rl_clientid;
4462
	struct nfs4_stateowner *sop;
4463
	struct nfs4_lockowner *lo;
4464
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4465
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4466
	struct list_head matches;
4467
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4468
	__be32 status;
4469
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4470 4471 4472 4473 4474 4475 4476

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

	/* XXX check for lease expiration */

	status = nfserr_stale_clientid;
4477
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
4478 4479 4480 4481
		return status;

	nfs4_lock_state();

4482 4483
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4484

4485
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4486 4487
		if (sop->so_is_open_owner)
			continue;
4488 4489 4490 4491 4492 4493 4494 4495
		if (!same_owner_str(sop, owner, clid))
			continue;
		list_for_each_entry(stp, &sop->so_stateids,
				st_perstateowner) {
			lo = lockowner(sop);
			if (check_for_locks(stp->st_file, lo))
				goto out;
			list_add(&lo->lo_list, &matches);
L
Linus Torvalds 已提交
4496
		}
4497 4498 4499 4500 4501
	}
	/* Clients probably won't expect us to return with some (but not all)
	 * of the lockowner state released; so don't release any until all
	 * have been checked. */
	status = nfs_ok;
N
NeilBrown 已提交
4502
	while (!list_empty(&matches)) {
4503 4504
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4505 4506
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4507 4508
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4509 4510 4511 4512 4513 4514 4515
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4516
alloc_reclaim(void)
L
Linus Torvalds 已提交
4517
{
N
NeilBrown 已提交
4518
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4519 4520
}

4521
bool
4522
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4523
{
4524
	struct nfs4_client_reclaim *crp;
4525

4526
	crp = nfsd4_find_reclaim_client(name, nn);
4527
	return (crp && crp->cr_clp);
4528 4529
}

L
Linus Torvalds 已提交
4530 4531 4532
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4533
struct nfs4_client_reclaim *
4534
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4535 4536
{
	unsigned int strhashval;
4537
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4538

N
NeilBrown 已提交
4539 4540
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4541 4542 4543
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4544
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4545
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4546
		crp->cr_clp = NULL;
4547
		nn->reclaim_str_hashtbl_size++;
4548 4549
	}
	return crp;
L
Linus Torvalds 已提交
4550 4551
}

4552
void
4553
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4554 4555 4556
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4557
	nn->reclaim_str_hashtbl_size--;
4558 4559
}

4560
void
4561
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4562 4563 4564 4565 4566
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4567 4568
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4569
			                struct nfs4_client_reclaim, cr_strhash);
4570
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4571 4572
		}
	}
4573
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4574 4575 4576 4577
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4578
struct nfs4_client_reclaim *
4579
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4580 4581 4582 4583
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4584
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4585

4586
	strhashval = clientstr_hashval(recdir);
4587
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4588
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4589 4590 4591 4592 4593 4594 4595 4596 4597
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4598
__be32
4599
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4600
{
4601 4602 4603
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4604
	clp = find_confirmed_client(clid, sessions, nn);
4605 4606 4607 4608
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4609 4610
}

B
Bryan Schumaker 已提交
4611 4612 4613 4614 4615 4616
#ifdef CONFIG_NFSD_FAULT_INJECTION

void nfsd_forget_clients(u64 num)
{
	struct nfs4_client *clp, *next;
	int count = 0;
4617
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4618 4619

	nfs4_lock_state();
4620
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
B
Bryan Schumaker 已提交
4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639
		expire_client(clp);
		if (++count == num)
			break;
	}
	nfs4_unlock_state();

	printk(KERN_INFO "NFSD: Forgot %d clients", count);
}

static void release_lockowner_sop(struct nfs4_stateowner *sop)
{
	release_lockowner(lockowner(sop));
}

static void release_openowner_sop(struct nfs4_stateowner *sop)
{
	release_openowner(openowner(sop));
}

4640
static int nfsd_release_n_owners(u64 num, bool is_open_owner,
4641 4642
				void (*release_sop)(struct nfs4_stateowner *),
				struct nfsd_net *nn)
B
Bryan Schumaker 已提交
4643 4644 4645 4646
{
	int i, count = 0;
	struct nfs4_stateowner *sop, *next;

4647
	for (i = 0; i < OWNER_HASH_SIZE; i++) {
4648
		list_for_each_entry_safe(sop, next, &nn->ownerstr_hashtbl[i], so_strhash) {
4649 4650
			if (sop->so_is_open_owner != is_open_owner)
				continue;
B
Bryan Schumaker 已提交
4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661
			release_sop(sop);
			if (++count == num)
				return count;
		}
	}
	return count;
}

void nfsd_forget_locks(u64 num)
{
	int count;
4662
	struct nfsd_net *nn = net_generic(&init_net, nfsd_net_id);
B
Bryan Schumaker 已提交
4663 4664

	nfs4_lock_state();
4665
	count = nfsd_release_n_owners(num, false, release_lockowner_sop, nn);
B
Bryan Schumaker 已提交
4666 4667 4668 4669 4670 4671 4672 4673
	nfs4_unlock_state();

	printk(KERN_INFO "NFSD: Forgot %d locks", count);
}

void nfsd_forget_openowners(u64 num)
{
	int count;
4674
	struct nfsd_net *nn = net_generic(&init_net, nfsd_net_id);
B
Bryan Schumaker 已提交
4675 4676

	nfs4_lock_state();
4677
	count = nfsd_release_n_owners(num, true, release_openowner_sop, nn);
B
Bryan Schumaker 已提交
4678 4679 4680 4681 4682
	nfs4_unlock_state();

	printk(KERN_INFO "NFSD: Forgot %d open owners", count);
}

4683
static int nfsd_process_n_delegations(u64 num, struct list_head *list)
B
Bryan Schumaker 已提交
4684 4685
{
	int i, count = 0;
4686 4687
	struct nfs4_file *fp, *fnext;
	struct nfs4_delegation *dp, *dnext;
B
Bryan Schumaker 已提交
4688 4689

	for (i = 0; i < FILE_HASH_SIZE; i++) {
4690 4691
		list_for_each_entry_safe(fp, fnext, &file_hashtbl[i], fi_hash) {
			list_for_each_entry_safe(dp, dnext, &fp->fi_delegations, dl_perfile) {
4692
				list_move(&dp->dl_recall_lru, list);
B
Bryan Schumaker 已提交
4693 4694 4695 4696 4697
				if (++count == num)
					return count;
			}
		}
	}
4698

B
Bryan Schumaker 已提交
4699 4700 4701 4702 4703 4704
	return count;
}

void nfsd_forget_delegations(u64 num)
{
	unsigned int count;
4705 4706 4707 4708 4709 4710
	LIST_HEAD(victims);
	struct nfs4_delegation *dp, *dnext;

	spin_lock(&recall_lock);
	count = nfsd_process_n_delegations(num, &victims);
	spin_unlock(&recall_lock);
B
Bryan Schumaker 已提交
4711 4712

	nfs4_lock_state();
4713 4714
	list_for_each_entry_safe(dp, dnext, &victims, dl_recall_lru)
		unhash_delegation(dp);
B
Bryan Schumaker 已提交
4715 4716 4717 4718 4719 4720 4721 4722
	nfs4_unlock_state();

	printk(KERN_INFO "NFSD: Forgot %d delegations", count);
}

void nfsd_recall_delegations(u64 num)
{
	unsigned int count;
4723 4724
	LIST_HEAD(victims);
	struct nfs4_delegation *dp, *dnext;
B
Bryan Schumaker 已提交
4725 4726

	spin_lock(&recall_lock);
4727 4728 4729 4730 4731
	count = nfsd_process_n_delegations(num, &victims);
	list_for_each_entry_safe(dp, dnext, &victims, dl_recall_lru) {
		list_del(&dp->dl_recall_lru);
		nfsd_break_one_deleg(dp);
	}
B
Bryan Schumaker 已提交
4732 4733 4734 4735 4736 4737 4738
	spin_unlock(&recall_lock);

	printk(KERN_INFO "NFSD: Recalled %d delegations", count);
}

#endif /* CONFIG_NFSD_FAULT_INJECTION */

4739
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4740

4741
void
4742
nfs4_state_init(void)
L
Linus Torvalds 已提交
4743
{
4744
	int i;
L
Linus Torvalds 已提交
4745 4746 4747 4748 4749

	for (i = 0; i < FILE_HASH_SIZE; i++) {
		INIT_LIST_HEAD(&file_hashtbl[i]);
	}
	INIT_LIST_HEAD(&del_recall_lru);
4750 4751
}

4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772
/*
 * 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);
}

4773 4774 4775 4776 4777 4778 4779 4780
static int nfs4_state_start_net(struct net *net)
{
	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)
4781
		goto err;
4782 4783 4784 4785
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4786 4787 4788 4789
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4790 4791 4792 4793
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4794 4795 4796 4797
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4798

4799
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4800
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4801
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4802
	}
4803 4804
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4805 4806
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4807 4808
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4809
	nn->conf_name_tree = RB_ROOT;
4810
	nn->unconf_name_tree = RB_ROOT;
4811
	INIT_LIST_HEAD(&nn->client_lru);
4812
	INIT_LIST_HEAD(&nn->close_lru);
4813
	spin_lock_init(&nn->client_lock);
4814

4815 4816
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);

4817
	return 0;
4818

4819 4820
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4821 4822
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4823 4824
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4825 4826
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4827 4828
err:
	return -ENOMEM;
4829 4830 4831 4832 4833 4834 4835 4836
}

static void
__nfs4_state_shutdown_net(struct net *net)
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4837
	struct rb_node *node, *tmp;
4838 4839 4840 4841 4842 4843 4844

	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);
		}
	}
4845 4846 4847 4848 4849 4850 4851 4852 4853 4854

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

4855
	kfree(nn->sessionid_hashtbl);
4856
	kfree(nn->lockowner_ino_hashtbl);
4857
	kfree(nn->ownerstr_hashtbl);
4858
	kfree(nn->unconf_id_hashtbl);
4859 4860 4861
	kfree(nn->conf_id_hashtbl);
}

4862 4863
/* initialization to perform when the nfsd service is started: */

4864 4865
int
nfs4_state_start(void)
4866
{
4867 4868
	struct net *net = &init_net;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4869 4870
	int ret;

4871 4872 4873 4874 4875 4876 4877
	/*
	 * FIXME: For now, we hang most of the pernet global stuff off of
	 * init_net until nfsd is fully containerized. Eventually, we'll
	 * need to pass a net pointer into this function, take a reference
	 * to that instead and then do most of the rest of this on a per-net
	 * basis.
	 */
4878
	get_net(net);
4879 4880 4881
	ret = nfs4_state_start_net(net);
	if (ret)
		return ret;
4882
	nfsd4_client_tracking_init(net);
4883
	nn->boot_time = get_seconds();
4884
	locks_start_grace(net, &nn->nfsd4_manager);
4885
	nn->grace_ended = false;
4886
	printk(KERN_INFO "NFSD: starting %ld-second grace period\n",
4887
	       nfsd4_grace);
4888
	ret = set_callback_cred();
4889 4890 4891 4892
	if (ret) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4893
	laundry_wq = create_singlethread_workqueue("nfsd4");
4894 4895 4896 4897
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4898 4899 4900
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4901 4902

	queue_delayed_work(laundry_wq, &nn->laundromat_work, nfsd4_grace * HZ);
4903
	set_max_delegations();
4904 4905 4906
	return 0;
out_free_laundry:
	destroy_workqueue(laundry_wq);
4907
out_recovery:
4908
	nfsd4_client_tracking_exit(net);
4909
	__nfs4_state_shutdown_net(net);
4910
	put_net(net);
4911
	return ret;
L
Linus Torvalds 已提交
4912 4913
}

4914
/* should be called with the state lock held */
L
Linus Torvalds 已提交
4915
static void
4916
__nfs4_state_shutdown(struct net *net)
L
Linus Torvalds 已提交
4917 4918 4919 4920
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;

4921
	__nfs4_state_shutdown_net(net);
4922

L
Linus Torvalds 已提交
4923 4924 4925 4926
	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
	list_for_each_safe(pos, next, &del_recall_lru) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
4927 4928
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4929 4930 4931 4932 4933 4934 4935 4936
		list_move(&dp->dl_recall_lru, &reaplist);
	}
	spin_unlock(&recall_lock);
	list_for_each_safe(pos, next, &reaplist) {
		dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
		unhash_delegation(dp);
	}

4937 4938
	nfsd4_client_tracking_exit(net);
	put_net(net);
L
Linus Torvalds 已提交
4939 4940 4941 4942 4943
}

void
nfs4_state_shutdown(void)
{
4944 4945 4946
	struct net *net = &init_net;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

4947
	cancel_delayed_work_sync(&nn->laundromat_work);
4948
	destroy_workqueue(laundry_wq);
4949
	locks_end_grace(&nn->nfsd4_manager);
4950
	__nfs4_state_shutdown(net);
4951
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
4952
}
4953 4954 4955 4956

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4957 4958
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
4959 4960 4961 4962 4963
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4964 4965 4966 4967 4968 4969 4970 4971 4972 4973
	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);
4974 4975
}

4976 4977 4978
/*
 * functions to set current state id
 */
4979 4980 4981 4982 4983 4984
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

4985 4986 4987 4988 4989 4990
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005
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
 */
5006

5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018
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);
}

5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030
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);
}

5031 5032 5033 5034 5035 5036 5037
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5038
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5039
{
5040
	get_stateid(cstate, &locku->lu_stateid);
5041
}
5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053

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);
}