nfs4state.c 132.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
/*
*  Copyright (c) 2001 The Regents of the University of Michigan.
*  All rights reserved.
*
*  Kendrick Smith <kmsmith@umich.edu>
*  Andy Adamson <kandros@umich.edu>
*
*  Redistribution and use in source and binary forms, with or without
*  modification, are permitted provided that the following conditions
*  are met:
*
*  1. Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*  2. Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in the
*     documentation and/or other materials provided with the distribution.
*  3. Neither the name of the University nor the names of its
*     contributors may be used to endorse or promote products derived
*     from this software without specific prior written permission.
*
*  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
*  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
*  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
*  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
*  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
*  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
*  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
*  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
*  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
*  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
*  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/

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

48 49
#include "netns.h"

L
Linus Torvalds 已提交
50 51
#define NFSDDBG_FACILITY                NFSDDBG_PROC

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

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

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

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

73 74 75
/* Locking: */

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

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

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

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

97 98 99 100 101 102 103 104 105 106 107 108 109
static void free_session(struct kref *);

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

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

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

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

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

static struct list_head del_recall_lru;

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

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

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

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

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

160 161 162 163
/* hash tables for lock and open owners */
#define OWNER_HASH_BITS              8
#define OWNER_HASH_SIZE             (1 << OWNER_HASH_BITS)
#define OWNER_HASH_MASK             (OWNER_HASH_SIZE - 1)
164

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

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

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

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

184 185
static struct list_head file_hashtbl[FILE_HASH_SIZE];

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

192 193 194 195 196 197 198 199 200 201
static void nfs4_file_get_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_get_access(fp, O_RDONLY);
		__nfs4_file_get_access(fp, O_WRONLY);
	} else
		__nfs4_file_get_access(fp, oflag);
}

static void nfs4_file_put_fd(struct nfs4_file *fp, int oflag)
202 203 204 205 206 207 208
{
	if (fp->fi_fds[oflag]) {
		fput(fp->fi_fds[oflag]);
		fp->fi_fds[oflag] = NULL;
	}
}

209
static void __nfs4_file_put_access(struct nfs4_file *fp, int oflag)
210 211 212
{
	if (atomic_dec_and_test(&fp->fi_access[oflag])) {
		nfs4_file_put_fd(fp, oflag);
213 214 215 216 217 218 219 220
		/*
		 * It's also safe to get rid of the RDWR open *if*
		 * we no longer have need of the other kind of access
		 * or if we already have the other kind of open:
		 */
		if (fp->fi_fds[1-oflag]
			|| atomic_read(&fp->fi_access[1 - oflag]) == 0)
			nfs4_file_put_fd(fp, O_RDWR);
221 222 223
	}
}

224 225 226 227 228 229 230 231 232
static void nfs4_file_put_access(struct nfs4_file *fp, int oflag)
{
	if (oflag == O_RDWR) {
		__nfs4_file_put_access(fp, O_RDONLY);
		__nfs4_file_put_access(fp, O_WRONLY);
	} else
		__nfs4_file_put_access(fp, oflag);
}

J
J. Bruce Fields 已提交
233
static inline int get_new_stid(struct nfs4_stid *stid)
234
{
J
J. Bruce Fields 已提交
235
	static int min_stateid = 0;
236
	struct idr *stateids = &stid->sc_client->cl_stateids;
J
J. Bruce Fields 已提交
237 238 239
	int new_stid;
	int error;

240
	error = idr_get_new_above(stateids, stid, min_stateid, &new_stid);
J
J. Bruce Fields 已提交
241
	/*
242 243 244 245
	 * 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
246
	 */
247
	WARN_ON_ONCE(error);
J
J. Bruce Fields 已提交
248 249 250 251 252 253 254 255 256
	/*
	 * 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):
	 */
257

J
J. Bruce Fields 已提交
258 259 260 261
	min_stateid = new_stid+1;
	if (min_stateid == INT_MAX)
		min_stateid = 0;
	return new_stid;
262 263
}

264
static void init_stid(struct nfs4_stid *stid, struct nfs4_client *cl, unsigned char type)
265 266
{
	stateid_t *s = &stid->sc_stateid;
J
J. Bruce Fields 已提交
267
	int new_id;
268 269 270 271

	stid->sc_type = type;
	stid->sc_client = cl;
	s->si_opaque.so_clid = cl->cl_clientid;
J
J. Bruce Fields 已提交
272 273
	new_id = get_new_stid(stid);
	s->si_opaque.so_id = (u32)new_id;
274 275
	/* Will be incremented before return to client: */
	s->si_generation = 0;
276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
}

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);
291 292
}

293 294 295 296 297
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
298
static struct nfs4_delegation *
299
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
300 301 302 303 304
{
	struct nfs4_delegation *dp;
	struct nfs4_file *fp = stp->st_file;

	dprintk("NFSD alloc_init_deleg\n");
305 306 307 308 309 310 311
	/*
	 * 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;
312 313
	if (fp->fi_had_conflict)
		return NULL;
314
	if (num_delegations > max_delegations)
315
		return NULL;
316
	dp = delegstateid(nfs4_alloc_stid(clp, deleg_slab));
N
NeilBrown 已提交
317
	if (dp == NULL)
L
Linus Torvalds 已提交
318
		return dp;
319
	init_stid(&dp->dl_stid, clp, NFS4_DELEG_STID);
320 321
	/*
	 * delegation seqid's are never incremented.  The 4.1 special
J
J. Bruce Fields 已提交
322 323
	 * meaning of seqid 0 isn't meaningful, really, but let's avoid
	 * 0 anyway just for consistency and use 1:
324 325
	 */
	dp->dl_stid.sc_stateid.si_generation = 1;
326
	num_delegations++;
327 328
	INIT_LIST_HEAD(&dp->dl_perfile);
	INIT_LIST_HEAD(&dp->dl_perclnt);
L
Linus Torvalds 已提交
329
	INIT_LIST_HEAD(&dp->dl_recall_lru);
330
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
331 332
	dp->dl_file = fp;
	dp->dl_type = type;
333
	fh_copy_shallow(&dp->dl_fh, &current_fh->fh_handle);
L
Linus Torvalds 已提交
334 335
	dp->dl_time = 0;
	atomic_set(&dp->dl_count, 1);
336
	nfsd4_init_callback(&dp->dl_recall);
L
Linus Torvalds 已提交
337 338 339 340 341 342 343 344
	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);
345
		put_nfs4_file(dp->dl_file);
N
NeilBrown 已提交
346
		kmem_cache_free(deleg_slab, dp);
347
		num_delegations--;
L
Linus Torvalds 已提交
348 349 350
	}
}

351
static void nfs4_put_deleg_lease(struct nfs4_file *fp)
L
Linus Torvalds 已提交
352
{
353 354 355
	if (atomic_dec_and_test(&fp->fi_delegees)) {
		vfs_setlease(fp->fi_deleg_file, F_UNLCK, &fp->fi_lease);
		fp->fi_lease = NULL;
356
		fput(fp->fi_deleg_file);
357 358
		fp->fi_deleg_file = NULL;
	}
L
Linus Torvalds 已提交
359 360
}

J
J. Bruce Fields 已提交
361 362
static void unhash_stid(struct nfs4_stid *s)
{
363 364 365
	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
366 367
}

L
Linus Torvalds 已提交
368 369 370 371
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
J
J. Bruce Fields 已提交
372
	unhash_stid(&dp->dl_stid);
373
	list_del_init(&dp->dl_perclnt);
L
Linus Torvalds 已提交
374
	spin_lock(&recall_lock);
375
	list_del_init(&dp->dl_perfile);
L
Linus Torvalds 已提交
376 377
	list_del_init(&dp->dl_recall_lru);
	spin_unlock(&recall_lock);
378
	nfs4_put_deleg_lease(dp->dl_file);
L
Linus Torvalds 已提交
379 380 381 382 383 384 385
	nfs4_put_delegation(dp);
}

/* 
 * SETCLIENTID state 
 */

386 387 388 389 390 391 392 393 394 395
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;
}

396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
/*
 * 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.
 */
414 415
static unsigned int
bmap_to_share_mode(unsigned long bmap) {
416
	int i;
417
	unsigned int access = 0;
418 419 420

	for (i = 1; i < 4; i++) {
		if (test_bit(i, &bmap))
421
			access |= i;
422
	}
423
	return access;
424 425
}

426
static bool
427
test_share(struct nfs4_ol_stateid *stp, struct nfsd4_open *open) {
428 429
	unsigned int access, deny;

430 431
	access = bmap_to_share_mode(stp->st_access_bmap);
	deny = bmap_to_share_mode(stp->st_deny_bmap);
432
	if ((access & open->op_share_deny) || (deny & open->op_share_access))
433 434
		return false;
	return true;
435 436
}

437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
/* 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);
}

458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
/* 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);
477 478 479 480
}

static int nfs4_access_to_omode(u32 access)
{
481
	switch (access & NFS4_SHARE_ACCESS_BOTH) {
482 483 484 485 486 487 488
	case NFS4_SHARE_ACCESS_READ:
		return O_RDONLY;
	case NFS4_SHARE_ACCESS_WRITE:
		return O_WRONLY;
	case NFS4_SHARE_ACCESS_BOTH:
		return O_RDWR;
	}
489 490
	WARN_ON_ONCE(1);
	return O_RDONLY;
491 492
}

493 494 495 496 497 498 499 500 501 502 503 504 505 506
/* 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);
	}
}

507
static void unhash_generic_stateid(struct nfs4_ol_stateid *stp)
508 509 510 511 512
{
	list_del(&stp->st_perfile);
	list_del(&stp->st_perstateowner);
}

513
static void close_generic_stateid(struct nfs4_ol_stateid *stp)
514
{
515
	release_all_access(stp);
O
OGAWA Hirofumi 已提交
516
	put_nfs4_file(stp->st_file);
517 518 519
	stp->st_file = NULL;
}

520
static void free_generic_stateid(struct nfs4_ol_stateid *stp)
521
{
522 523 524
	kmem_cache_free(stateid_slab, stp);
}

525
static void release_lock_stateid(struct nfs4_ol_stateid *stp)
526 527 528 529
{
	struct file *file;

	unhash_generic_stateid(stp);
J
J. Bruce Fields 已提交
530
	unhash_stid(&stp->st_stid);
531 532
	file = find_any_file(stp->st_file);
	if (file)
533
		locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
534
	close_generic_stateid(stp);
535 536 537
	free_generic_stateid(stp);
}

538
static void unhash_lockowner(struct nfs4_lockowner *lo)
539
{
540
	struct nfs4_ol_stateid *stp;
541

542 543
	list_del(&lo->lo_owner.so_strhash);
	list_del(&lo->lo_perstateid);
544
	list_del(&lo->lo_owner_ino_hash);
545 546
	while (!list_empty(&lo->lo_owner.so_stateids)) {
		stp = list_first_entry(&lo->lo_owner.so_stateids,
547
				struct nfs4_ol_stateid, st_perstateowner);
548 549 550 551
		release_lock_stateid(stp);
	}
}

552
static void release_lockowner(struct nfs4_lockowner *lo)
553
{
554 555
	unhash_lockowner(lo);
	nfs4_free_lockowner(lo);
556 557 558
}

static void
559
release_stateid_lockowners(struct nfs4_ol_stateid *open_stp)
560
{
561
	struct nfs4_lockowner *lo;
562 563

	while (!list_empty(&open_stp->st_lockowners)) {
564 565 566
		lo = list_entry(open_stp->st_lockowners.next,
				struct nfs4_lockowner, lo_perstateid);
		release_lockowner(lo);
567 568 569
	}
}

570
static void unhash_open_stateid(struct nfs4_ol_stateid *stp)
571 572 573
{
	unhash_generic_stateid(stp);
	release_stateid_lockowners(stp);
574 575 576 577 578 579
	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
580
	unhash_stid(&stp->st_stid);
581 582 583
	free_generic_stateid(stp);
}

584
static void unhash_openowner(struct nfs4_openowner *oo)
585
{
586
	struct nfs4_ol_stateid *stp;
587

588 589 590 591
	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,
592
				struct nfs4_ol_stateid, st_perstateowner);
593
		release_open_stateid(stp);
594 595 596
	}
}

597 598 599 600 601
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
602
		unhash_stid(&s->st_stid);
603 604 605 606 607
		free_generic_stateid(s);
		oo->oo_last_closed_stid = NULL;
	}
}

608
static void release_openowner(struct nfs4_openowner *oo)
609
{
610 611
	unhash_openowner(oo);
	list_del(&oo->oo_close_lru);
612
	release_last_closed_stateid(oo);
613
	nfs4_free_openowner(oo);
614 615
}

M
Marc Eshel 已提交
616 617 618 619 620 621 622 623
static inline int
hash_sessionid(struct nfs4_sessionid *sessionid)
{
	struct nfsd4_sessionid *sid = (struct nfsd4_sessionid *)sessionid;

	return sid->sequence % SESSION_HASH_SIZE;
}

624
#ifdef NFSD_DEBUG
M
Marc Eshel 已提交
625 626 627 628 629 630
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]);
}
631 632 633 634 635 636 637
#else
static inline void
dump_sessionid(const char *fn, struct nfs4_sessionid *sessionid)
{
}
#endif

M
Marc Eshel 已提交
638

A
Andy Adamson 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651
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;
}

/*
652 653 654 655 656 657 658 659 660 661 662 663 664
 * 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)

665 666 667 668 669 670 671 672 673
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]);
}

674
/*
675 676 677 678 679 680 681 682
 * 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;
}

683
static int nfsd4_sanitize_slot_size(u32 size)
A
Andy Adamson 已提交
684
{
685 686
	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
687

688 689
	return size;
}
A
Andy Adamson 已提交
690

691 692
/*
 * XXX: If we run out of reserved DRC memory we could (up to a point)
693
 * re-negotiate active sessions and reduce their slot usage to make
694
 * room for new connections. For now we just fail the create session.
A
Andy Adamson 已提交
695
 */
696
static int nfsd4_get_drc_mem(int slotsize, u32 num)
A
Andy Adamson 已提交
697
{
698
	int avail;
A
Andy Adamson 已提交
699

700
	num = min_t(u32, num, NFSD_MAX_SLOTS_PER_SESSION);
701

702 703 704 705 706 707
	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
708

709 710
	return num;
}
A
Andy Adamson 已提交
711

712 713
static void nfsd4_put_drc_mem(int slotsize, int num)
{
714
	spin_lock(&nfsd_drc_lock);
715
	nfsd_drc_mem_used -= slotsize * num;
716
	spin_unlock(&nfsd_drc_lock);
717
}
A
Andy Adamson 已提交
718

719
static struct nfsd4_session *__alloc_session(int slotsize, int numslots)
720 721 722
{
	struct nfsd4_session *new;
	int mem, i;
723

724 725 726
	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
727

728 729 730
	new = kzalloc(sizeof(*new) + mem, GFP_KERNEL);
	if (!new)
		return NULL;
731
	/* allocate each struct nfsd4_slot and data cache in one piece */
732 733 734 735
	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])
736 737
			goto out_free;
	}
738 739 740 741 742 743
	return new;
out_free:
	while (i--)
		kfree(new->se_slots[i]);
	kfree(new);
	return NULL;
A
Andy Adamson 已提交
744 745
}

746
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
747
{
748
	u32 maxrpc = nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
749

750
	new->maxreqs = numslots;
751 752
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
753 754 755 756
	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
757

758 759 760 761 762
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
763

764 765 766 767
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
768

769 770 771 772 773 774
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
	spin_unlock(&clp->cl_lock);
775
	nfsd4_probe_callback(clp);
776
}
A
Andy Adamson 已提交
777

778
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
779 780
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
781

782 783
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
784
		return NULL;
785 786
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
787
	conn->cn_flags = flags;
788 789 790
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
791

792 793 794 795
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
796 797
}

798
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
799
{
800
	struct nfs4_client *clp = ses->se_client;
801

802
	spin_lock(&clp->cl_lock);
803
	__nfsd4_hash_conn(conn, ses);
804
	spin_unlock(&clp->cl_lock);
805 806
}

807
static int nfsd4_register_conn(struct nfsd4_conn *conn)
808
{
809
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
810
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
811 812
}

813
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
814
{
815
	int ret;
A
Andy Adamson 已提交
816

817
	nfsd4_hash_conn(conn, ses);
818 819 820 821
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
822
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
823 824 825
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
826
}
A
Andy Adamson 已提交
827

828
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
829 830 831
{
	u32 dir = NFS4_CDFC4_FORE;

832
	if (cses->flags & SESSION4_BACK_CHAN)
833
		dir |= NFS4_CDFC4_BACK;
834
	return alloc_conn(rqstp, dir);
835 836 837
}

/* must be called under client_lock */
838
static void nfsd4_del_conns(struct nfsd4_session *s)
839
{
840 841
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
842

843 844 845 846 847
	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);
848

849 850
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
851

852 853 854
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
855
}
A
Andy Adamson 已提交
856

857 858 859 860 861 862 863
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);
}

864
static void free_session(struct kref *kref)
865 866
{
	struct nfsd4_session *ses;
867
	struct nfsd_net *nn;
868 869

	ses = container_of(kref, struct nfsd4_session, se_ref);
870 871 872
	nn = net_generic(ses->se_client->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
873
	nfsd4_del_conns(ses);
874
	__free_session(ses);
875 876
}

877 878
void nfsd4_put_session(struct nfsd4_session *ses)
{
879 880 881
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
882
	nfsd4_put_session_locked(ses);
883
	spin_unlock(&nn->client_lock);
884 885
}

886
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan)
887 888 889 890 891 892 893 894 895 896 897 898
{
	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);
899 900
	if (numslots < 1)
		return NULL;
901

902
	new = __alloc_session(slotsize, numslots);
903 904
	if (!new) {
		nfsd4_put_drc_mem(slotsize, fchan->maxreqs);
905
		return NULL;
906
	}
907
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize);
908 909
	return new;
}
910

911
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
912 913
{
	int idx;
914
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
915

A
Andy Adamson 已提交
916 917 918
	new->se_client = clp;
	gen_sessionid(new);

919 920
	INIT_LIST_HEAD(&new->se_conns);

921
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
922
	new->se_flags = cses->flags;
923
	new->se_cb_prog = cses->callback_prog;
924
	new->se_cb_sec = cses->cb_sec;
A
Andy Adamson 已提交
925
	kref_init(&new->se_ref);
926
	idx = hash_sessionid(&new->se_sessionid);
927
	spin_lock(&nn->client_lock);
928
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
929
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
930
	list_add(&new->se_perclnt, &clp->cl_sessions);
931
	spin_unlock(&clp->cl_lock);
932
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
933

934
	if (cses->flags & SESSION4_BACK_CHAN) {
935
		struct sockaddr *sa = svc_addr(rqstp);
936 937 938 939 940 941 942
		/*
		 * 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:
		 */
943 944 945
		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
946 947
}

948
/* caller must hold client_lock */
M
Marc Eshel 已提交
949
static struct nfsd4_session *
950
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
951 952 953
{
	struct nfsd4_session *elem;
	int idx;
954
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
955 956 957 958

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
959
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
960 961 962 963 964 965 966 967 968 969
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

970
/* caller must hold client_lock */
A
Andy Adamson 已提交
971
static void
M
Marc Eshel 已提交
972
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
973 974
{
	list_del(&ses->se_hash);
975
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
976
	list_del(&ses->se_perclnt);
977
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
978 979
}

980
/* must be called under the client_lock */
L
Linus Torvalds 已提交
981
static inline void
982
renew_client_locked(struct nfs4_client *clp)
L
Linus Torvalds 已提交
983
{
984 985
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

B
Benny Halevy 已提交
986
	if (is_client_expired(clp)) {
987 988
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
B
Benny Halevy 已提交
989 990 991 992 993 994
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

L
Linus Torvalds 已提交
995 996 997
	dprintk("renewing client (clientid %08x/%08x)\n", 
			clp->cl_clientid.cl_boot, 
			clp->cl_clientid.cl_id);
998
	list_move_tail(&clp->cl_lru, &nn->client_lru);
L
Linus Torvalds 已提交
999 1000 1001
	clp->cl_time = get_seconds();
}

1002 1003 1004
static inline void
renew_client(struct nfs4_client *clp)
{
1005 1006 1007
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
1008
	renew_client_locked(clp);
1009
	spin_unlock(&nn->client_lock);
1010 1011
}

L
Linus Torvalds 已提交
1012 1013
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1014
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1015
{
1016
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1017
		return 0;
A
Andy Adamson 已提交
1018
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1019
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027
	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.
 */
1028
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1029 1030 1031
{
	struct nfs4_client *clp;

1032 1033 1034
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1035
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1036 1037 1038
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1039
	}
1040
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1047 1048 1049
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
1050 1051 1052 1053 1054
	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);
1055
		nfsd4_put_session_locked(ses);
1056
	}
1057
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1058 1059 1060 1061
	kfree(clp->cl_name.data);
	kfree(clp);
}

1062 1063 1064 1065
void
release_session_client(struct nfsd4_session *session)
{
	struct nfs4_client *clp = session->se_client;
1066
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
1067

1068
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
1069 1070 1071 1072 1073 1074
		return;
	if (is_client_expired(clp)) {
		free_client(clp);
		session->se_client = NULL;
	} else
		renew_client_locked(clp);
1075
	spin_unlock(&nn->client_lock);
1076 1077
}

B
Benny Halevy 已提交
1078 1079 1080 1081
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1082 1083
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1084
	mark_client_expired(clp);
B
Benny Halevy 已提交
1085
	list_del(&clp->cl_lru);
1086
	spin_lock(&clp->cl_lock);
1087 1088
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1089
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1090 1091
}

L
Linus Torvalds 已提交
1092
static void
1093
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1094
{
1095
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1096 1097
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1098
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1099 1100 1101

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1102 1103 1104
	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
1105 1106 1107 1108 1109 1110 1111
		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);
	}
1112
	while (!list_empty(&clp->cl_openowners)) {
1113 1114
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1115
	}
1116
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1117 1118
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1119
	list_del(&clp->cl_idhash);
1120
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1121
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1122
	else
1123
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1124
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1125
	unhash_client_locked(clp);
1126 1127
	if (atomic_read(&clp->cl_refcount) == 0)
		free_client(clp);
1128
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1129 1130
}

1131 1132 1133 1134 1135 1136
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1137 1138 1139 1140
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
1141 1142
}

1143 1144
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1145 1146 1147 1148
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1149
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1150
{
1151 1152 1153 1154 1155 1156 1157
	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;
1158
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1159 1160 1161 1162
	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);
1163
	return 0;
L
Linus Torvalds 已提交
1164 1165
}

1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176
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);
}

1177
static int same_name(const char *n1, const char *n2)
1178
{
N
NeilBrown 已提交
1179
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1180 1181 1182
}

static int
1183 1184 1185
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1186 1187 1188
}

static int
1189 1190 1191
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
1192 1193
}

1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
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;
}

1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
/*
 * 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);
}


1222
static bool
1223 1224
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1225
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1226
		|| (cr1->cr_uid != cr2->cr_uid)
1227 1228 1229 1230 1231 1232 1233
		|| (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;
1234
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1235 1236
}

1237
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1238 1239 1240
{
	static u32 current_clientid = 1;

1241
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1242 1243 1244
	clp->cl_clientid.cl_id = current_clientid++; 
}

1245 1246
static void gen_confirm(struct nfs4_client *clp)
{
1247
	__be32 verf[2];
1248
	static u32 i;
L
Linus Torvalds 已提交
1249

1250 1251 1252
	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
1253 1254
}

1255
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1256
{
1257
	return idr_find(&cl->cl_stateids, t->si_opaque.so_id);
J
J. Bruce Fields 已提交
1258 1259
}

1260
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1261 1262
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1263

1264
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1265 1266
	if (!s)
		return NULL;
1267
	if (typemask & s->sc_type)
1268 1269 1270 1271
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1272
static struct nfs4_client *create_client(struct xdr_netobj name,
1273 1274 1275 1276
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1277
	int ret;
1278
	struct net *net = SVC_NET(rqstp);
1279
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1280 1281 1282 1283 1284

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

1285
	INIT_LIST_HEAD(&clp->cl_sessions);
1286 1287
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1288
		spin_lock(&nn->client_lock);
1289
		free_client(clp);
1290
		spin_unlock(&nn->client_lock);
1291
		return NULL;
1292
	}
1293
	idr_init(&clp->cl_stateids);
1294
	atomic_set(&clp->cl_refcount, 0);
1295
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1296 1297 1298 1299
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1300
	INIT_LIST_HEAD(&clp->cl_callbacks);
1301
	spin_lock_init(&clp->cl_lock);
1302
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1303
	clp->cl_time = get_seconds();
1304 1305 1306 1307 1308
	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);
1309
	clp->cl_cb_session = NULL;
1310
	clp->net = net;
1311 1312 1313
	return clp;
}

1314
static void
1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
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
1356 1357
{
	unsigned int idhashval;
1358
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1359

1360
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1361
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1362
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1363
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1364
	renew_client(clp);
L
Linus Torvalds 已提交
1365 1366
}

1367
static void
L
Linus Torvalds 已提交
1368 1369 1370
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1371
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1372 1373

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1374
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1375
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1376
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1377
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1378 1379 1380 1381
	renew_client(clp);
}

static struct nfs4_client *
1382
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1383 1384 1385 1386
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1387
	list_for_each_entry(clp, &nn->conf_id_hashtbl[idhashval], cl_idhash) {
1388
		if (same_clid(&clp->cl_clientid, clid)) {
1389 1390
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1391
			renew_client(clp);
L
Linus Torvalds 已提交
1392
			return clp;
1393
		}
L
Linus Torvalds 已提交
1394 1395 1396 1397 1398
	}
	return NULL;
}

static struct nfs4_client *
1399
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1400 1401 1402 1403
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1404
	list_for_each_entry(clp, &nn->unconf_id_hashtbl[idhashval], cl_idhash) {
1405 1406 1407
		if (same_clid(&clp->cl_clientid, clid)) {
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
L
Linus Torvalds 已提交
1408
			return clp;
1409
		}
L
Linus Torvalds 已提交
1410 1411 1412 1413
	}
	return NULL;
}

1414
static bool clp_used_exchangeid(struct nfs4_client *clp)
1415
{
1416
	return clp->cl_exchange_flags != 0;
1417
} 
1418

1419
static struct nfs4_client *
1420
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1421
{
1422
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1423 1424 1425
}

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

1431
static void
1432
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1433
{
1434
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1435 1436
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
	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
1447 1448
		goto out_err;

1449
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1450
					    se->se_callback_addr_len,
1451 1452
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1453

1454
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1455
		goto out_err;
1456

1457 1458
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1459

1460 1461
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1462
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1463 1464
	return;
out_err:
1465 1466
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1467
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1468 1469 1470 1471 1472 1473
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1474
/*
1475
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1476 1477 1478 1479
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1480 1481
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1482

1483
	dprintk("--> %s slot %p\n", __func__, slot);
1484

1485 1486
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1487

1488
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1489
	if (nfsd4_not_cached(resp)) {
1490
		slot->sl_datalen = 0;
1491
		return;
1492
	}
1493 1494 1495 1496 1497 1498 1499
	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;
1500 1501 1502
}

/*
1503 1504 1505 1506
 * 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.
1507 1508
 *
 */
1509 1510 1511
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1512
{
1513 1514
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1515

1516 1517 1518
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1519

1520
	/* Return nfserr_retry_uncached_rep in next operation. */
1521
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1522 1523 1524
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1525
	}
1526
	return op->status;
1527 1528 1529
}

/*
1530 1531
 * The sequence operation is not cached because we can use the slot and
 * session values.
1532 1533
 */
__be32
1534 1535
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1536
{
1537
	struct nfsd4_slot *slot = resp->cstate.slot;
1538 1539
	__be32 status;

1540
	dprintk("--> %s slot %p\n", __func__, slot);
1541

1542 1543 1544 1545
	/* 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;
1546

1547 1548
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1549

1550 1551 1552
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1553 1554 1555 1556

	return status;
}

A
Andy Adamson 已提交
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
/*
 * 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;
}

1573 1574 1575 1576 1577 1578 1579 1580
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.
	 */
1581 1582 1583
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1584 1585
}

A
Andy Adamson 已提交
1586 1587 1588 1589 1590
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1591
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1592
	__be32 status;
1593
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1594
	nfs4_verifier		verf = exid->verifier;
1595
	struct sockaddr		*sa = svc_addr(rqstp);
1596
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1597
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1598

1599
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1600
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1601
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1602
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1603
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1604

1605
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1606 1607 1608 1609 1610 1611
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1612 1613
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1614 1615 1616 1617 1618
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1619
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1620
	nfs4_lock_state();
1621
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1622
	if (conf) {
1623 1624 1625
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1626 1627
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1628
				status = nfserr_inval;
1629 1630
				goto out;
			}
1631
			if (!creds_match) { /* case 9 */
1632
				status = nfserr_perm;
1633 1634 1635
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1636 1637 1638
				status = nfserr_not_same;
				goto out;
			}
1639 1640 1641 1642
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1643
		}
1644
		if (!creds_match) { /* case 3 */
1645 1646
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1647 1648 1649 1650 1651
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1652
		if (verfs_match) { /* case 2 */
1653
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1654 1655 1656 1657 1658
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1659 1660
	}

1661
	if (update) { /* case 7 */
1662 1663
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1664 1665
	}

1666
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1667
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1668 1669
		expire_client(unconf);

1670
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1671
out_new:
J
Jeff Layton 已提交
1672
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1673
	if (new == NULL) {
1674
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1675 1676
		goto out;
	}
1677
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1678

1679
	gen_clid(new, nn);
1680
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1681 1682 1683 1684
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1685
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1686 1687 1688
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1689
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1690 1691 1692 1693 1694
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1695 1696
}

J
J. Bruce Fields 已提交
1697
static __be32
1698
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1699
{
1700 1701
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1702 1703

	/* The slot is in use, and no response has been sent. */
1704 1705
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1706 1707 1708 1709
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1710
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1711
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1712
		return nfs_ok;
1713
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1714 1715 1716 1717
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1718 1719 1720 1721 1722 1723 1724
/*
 * 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
1725
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
{
	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;
}

1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
#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
1756
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1757 1758 1759 1760 1761
{
	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
1762 1763 1764 1765 1766
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1767
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1768
	struct nfs4_client *conf, *unconf;
1769
	struct nfsd4_session *new;
1770
	struct nfsd4_conn *conn;
1771
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1772
	__be32 status = 0;
1773
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1774

1775 1776
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1777 1778
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1779 1780 1781 1782 1783 1784 1785
	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;
1786

A
Andy Adamson 已提交
1787
	nfs4_lock_state();
1788
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1789
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
A
Andy Adamson 已提交
1790 1791

	if (conf) {
1792 1793
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1794
		if (status == nfserr_replay_cache) {
1795
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1796
			goto out_free_conn;
1797
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1798
			status = nfserr_seq_misordered;
1799
			goto out_free_conn;
A
Andy Adamson 已提交
1800 1801
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1802
		struct nfs4_client *old;
A
Andy Adamson 已提交
1803
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1804
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1805
			status = nfserr_clid_inuse;
1806
			goto out_free_conn;
A
Andy Adamson 已提交
1807
		}
1808 1809
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1810 1811
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1812
			status = nfserr_seq_misordered;
1813
			goto out_free_conn;
A
Andy Adamson 已提交
1814
		}
1815
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
J
J. Bruce Fields 已提交
1816 1817 1818
		if (old)
			expire_client(old);
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1819 1820 1821
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1822
		goto out_free_conn;
A
Andy Adamson 已提交
1823
	}
1824
	status = nfs_ok;
1825 1826 1827 1828 1829 1830
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1831 1832 1833
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1834
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1835
	       NFS4_MAX_SESSIONID_LEN);
1836 1837
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1838
	cs_slot->sl_seqid++;
1839
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1840

1841 1842
	/* 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
1843 1844 1845 1846
out:
	nfs4_unlock_state();
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
1847 1848 1849 1850 1851
out_free_conn:
	free_conn(conn);
out_free_session:
	__free_session(new);
	goto out;
A
Andy Adamson 已提交
1852 1853
}

1854 1855 1856 1857 1858 1859 1860 1861
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;
}

1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
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;
}

1876 1877 1878
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1879
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1880

1881
	spin_lock(&nn->client_lock);
1882 1883
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1884
	spin_unlock(&nn->client_lock);
1885 1886 1887 1888 1889 1890

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1891 1892 1893 1894 1895
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1896
	struct nfsd4_conn *conn;
1897
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1898 1899 1900

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1901
	spin_lock(&nn->client_lock);
1902
	cstate->session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1903 1904 1905 1906 1907 1908
	/* 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);
	}
1909
	spin_unlock(&nn->client_lock);
1910 1911 1912 1913
	if (!cstate->session)
		return nfserr_badsession;

	status = nfsd4_map_bcts_dir(&bcts->dir);
1914 1915 1916 1917 1918
	if (status)
		return status;
	conn = alloc_conn(rqstp, bcts->dir);
	if (!conn)
		return nfserr_jukebox;
1919
	nfsd4_init_conn(rqstp, conn, cstate->session);
1920
	return nfs_ok;
1921 1922
}

1923 1924 1925 1926 1927 1928 1929
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
1930 1931 1932 1933 1934
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1935
	struct nfsd4_session *ses;
J
J. Bruce Fields 已提交
1936
	__be32 status = nfserr_badsession;
1937
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1938 1939 1940 1941 1942 1943 1944 1945

	/* 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?
	 */

1946
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1947 1948 1949
		if (!nfsd4_last_compound_op(r))
			return nfserr_not_only_op;
	}
B
Benny Halevy 已提交
1950
	dump_sessionid(__func__, &sessionid->sessionid);
1951
	spin_lock(&nn->client_lock);
1952
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
B
Benny Halevy 已提交
1953
	if (!ses) {
1954
		spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1955 1956 1957 1958
		goto out;
	}

	unhash_session(ses);
1959
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1960

1961
	nfs4_lock_state();
1962
	nfsd4_probe_callback_sync(ses->se_client);
1963
	nfs4_unlock_state();
1964

1965
	spin_lock(&nn->client_lock);
1966
	nfsd4_del_conns(ses);
1967
	nfsd4_put_session_locked(ses);
1968
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1969 1970 1971 1972
	status = nfs_ok;
out:
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
1973 1974
}

1975
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
1976 1977 1978 1979
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
1980
		if (c->cn_xprt == xpt) {
1981 1982 1983 1984 1985 1986
			return c;
		}
	}
	return NULL;
}

1987
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
1988 1989
{
	struct nfs4_client *clp = ses->se_client;
1990
	struct nfsd4_conn *c;
1991
	int ret;
1992 1993

	spin_lock(&clp->cl_lock);
1994
	c = __nfsd4_find_conn(new->cn_xprt, ses);
1995 1996 1997 1998 1999 2000 2001
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
2002 2003 2004 2005
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2006 2007 2008
	return;
}

2009 2010 2011 2012 2013 2014 2015
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
2016 2017 2018 2019 2020 2021 2022 2023
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
2024
__be32
B
Benny Halevy 已提交
2025
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2026 2027 2028
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2029
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2030 2031
	struct nfsd4_session *session;
	struct nfsd4_slot *slot;
2032
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2033
	__be32 status;
2034
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2035

2036 2037 2038
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2039 2040 2041 2042 2043 2044 2045 2046
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2047
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2048
	status = nfserr_badsession;
2049
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2050 2051 2052
	if (!session)
		goto out;

2053 2054 2055 2056
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
		goto out;

M
Mi Jinlong 已提交
2057 2058 2059 2060
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
		goto out;

B
Benny Halevy 已提交
2061
	status = nfserr_badslot;
2062
	if (seq->slotid >= session->se_fchannel.maxreqs)
B
Benny Halevy 已提交
2063 2064
		goto out;

2065
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2066 2067
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2068 2069 2070 2071 2072
	/* 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;

2073 2074
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2075
	if (status == nfserr_replay_cache) {
2076 2077 2078
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
			goto out;
B
Benny Halevy 已提交
2079 2080
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2081
		/* Return the cached reply status and set cstate->status
2082
		 * for nfsd4_proc_compound processing */
2083
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2084
		cstate->status = nfserr_replay_cache;
2085
		goto out;
B
Benny Halevy 已提交
2086 2087 2088 2089
	}
	if (status)
		goto out;

2090 2091
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2092

B
Benny Halevy 已提交
2093 2094
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2095
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2096 2097
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2098 2099
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2100 2101 2102 2103 2104

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

out:
2105
	/* Hold a session reference until done processing the compound. */
2106
	if (cstate->session) {
2107 2108
		struct nfs4_client *clp = session->se_client;

2109
		nfsd4_get_session(cstate->session);
2110
		atomic_inc(&clp->cl_refcount);
2111 2112
		switch (clp->cl_cb_state) {
		case NFSD4_CB_DOWN:
2113
			seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
2114 2115
			break;
		case NFSD4_CB_FAULT:
2116
			seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
2117
			break;
2118 2119
		default:
			seq->status_flags = 0;
2120
		}
2121
	}
2122
	kfree(conn);
2123
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2124 2125
	dprintk("%s: return %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
2126 2127
}

2128 2129 2130 2131
__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
2132
	__be32 status = 0;
2133
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2134 2135

	nfs4_lock_state();
2136
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2137
	conf = find_confirmed_client(&dc->clientid, true, nn);
2138 2139 2140 2141

	if (conf) {
		clp = conf;

2142
		if (!is_client_expired(conf) && client_has_state(conf)) {
2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165
			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;
}

2166 2167 2168
__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
2169
	__be32 status = 0;
2170

2171 2172 2173 2174 2175 2176 2177 2178 2179
	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;
	}
2180

2181
	nfs4_lock_state();
2182
	status = nfserr_complete_already;
2183 2184
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2185 2186 2187 2188
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2189 2190 2191 2192 2193 2194 2195
		/*
		 * 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.
		 */
2196 2197 2198
		goto out;

	status = nfs_ok;
2199
	nfsd4_client_record_create(cstate->session->se_client);
2200
out:
2201
	nfs4_unlock_state();
2202
	return status;
2203 2204
}

2205
__be32
2206 2207
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2208
{
2209
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2210
	nfs4_verifier		clverifier = setclid->se_verf;
2211
	struct nfs4_client	*conf, *unconf, *new;
2212
	__be32 			status;
2213 2214
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2215
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2216
	nfs4_lock_state();
2217
	conf = find_confirmed_client_by_name(&clname, nn);
2218
	if (conf) {
2219
		/* case 0: */
L
Linus Torvalds 已提交
2220
		status = nfserr_clid_inuse;
2221 2222
		if (clp_used_exchangeid(conf))
			goto out;
2223
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2224 2225 2226 2227 2228
			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
2229 2230 2231
			goto out;
		}
	}
2232
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2233 2234
	if (unconf)
		expire_client(unconf);
2235
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2236
	new = create_client(clname, rqstp, &clverifier);
2237 2238
	if (new == NULL)
		goto out;
2239
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2240
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2241
		copy_clid(new, conf);
2242
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2243
		gen_clid(new, nn);
2244
	new->cl_minorversion = 0;
2245
	gen_callback(new, setclid, rqstp);
2246
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
	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;
}


2257
__be32
2258 2259 2260
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2261
{
2262
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2263 2264
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2265
	__be32 status;
2266
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2267

2268
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2269 2270
		return nfserr_stale_clientid;
	nfs4_lock_state();
2271

2272
	conf = find_confirmed_client(clid, false, nn);
2273
	unconf = find_unconfirmed_client(clid, false, nn);
2274
	/*
2275 2276 2277 2278
	 * 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.
2279
	 */
2280 2281 2282 2283 2284
	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;
2285
	/* cases below refer to rfc 3530 section 14.2.34: */
2286 2287
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2288
			status = nfs_ok;
2289 2290 2291 2292 2293 2294
		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 */
2295 2296 2297
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2298
	} else { /* case 3: normal case; new or rebooted client */
2299
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2300
		if (conf)
2301 2302
			expire_client(conf);
		move_to_confirmed(unconf);
2303
		nfsd4_probe_callback(unconf);
2304
	}
L
Linus Torvalds 已提交
2305 2306 2307 2308 2309
out:
	nfs4_unlock_state();
	return status;
}

2310 2311 2312 2313 2314
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2315
/* OPEN Share state helper functions */
2316
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2317 2318 2319
{
	unsigned int hashval = file_hashval(ino);

2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331
	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
2332 2333
}

N
NeilBrown 已提交
2334
static void
2335
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2336
{
N
NeilBrown 已提交
2337 2338
	if (*slab == NULL)
		return;
2339
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2340
	*slab = NULL;
L
Linus Torvalds 已提交
2341 2342
}

2343
void
L
Linus Torvalds 已提交
2344 2345
nfsd4_free_slabs(void)
{
2346 2347
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2348
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2349
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2350
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2351
}
L
Linus Torvalds 已提交
2352

2353
int
N
NeilBrown 已提交
2354 2355
nfsd4_init_slabs(void)
{
2356 2357 2358 2359 2360
	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",
2361
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2362
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2363 2364
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2365
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2366 2367
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2368
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2369
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2370 2371
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2372
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2373
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2374 2375
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2376 2377 2378 2379 2380
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2381 2382
}

2383 2384 2385 2386 2387 2388 2389
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
2390
{
2391 2392
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2393 2394
}

2395
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2396
{
2397 2398 2399
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2400 2401
}

2402
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2403
{
L
Linus Torvalds 已提交
2404 2405
	struct nfs4_stateowner *sop;

2406
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2407 2408 2409 2410 2411
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2412
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2413
		return NULL;
2414 2415 2416
	}
	sop->so_owner.len = owner->len;

2417
	INIT_LIST_HEAD(&sop->so_stateids);
2418 2419 2420 2421 2422
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2423
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2424
{
2425 2426 2427
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2428
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2429 2430
}

2431
static struct nfs4_openowner *
2432
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2433
	struct nfs4_openowner *oo;
2434

2435 2436
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2437
		return NULL;
2438 2439
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2440
	oo->oo_flags = NFS4_OO_NEW;
2441
	oo->oo_time = 0;
2442
	oo->oo_last_closed_stid = NULL;
2443 2444 2445
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2446 2447
}

2448
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2449
	struct nfs4_openowner *oo = open->op_openowner;
2450
	struct nfs4_client *clp = oo->oo_owner.so_client;
L
Linus Torvalds 已提交
2451

2452
	init_stid(&stp->st_stid, clp, NFS4_OPEN_STID);
2453
	INIT_LIST_HEAD(&stp->st_lockowners);
2454
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2455
	list_add(&stp->st_perfile, &fp->fi_stateids);
2456
	stp->st_stateowner = &oo->oo_owner;
2457
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2458 2459 2460
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2461
	set_access(open->op_share_access, stp);
2462
	set_deny(open->op_share_deny, stp);
2463
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2464 2465
}

2466
static void
2467
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2468
{
2469 2470
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2471
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2472

2473
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2474
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2475 2476 2477
}

static int
2478 2479 2480 2481 2482 2483
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
2484 2485
}

2486
static struct nfs4_openowner *
2487 2488
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2489
{
2490 2491
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2492
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2493

2494
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2495 2496
		if (!so->so_is_open_owner)
			continue;
2497 2498
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2499 2500 2501
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2502 2503 2504
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515
	}
	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;

2516
	spin_lock(&recall_lock);
L
Linus Torvalds 已提交
2517
	list_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2518 2519
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2520
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2521
			return fp;
2522
		}
L
Linus Torvalds 已提交
2523
	}
2524
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2525 2526 2527 2528 2529 2530 2531
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2532
static __be32
L
Linus Torvalds 已提交
2533 2534 2535 2536
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;
2537
	struct nfs4_ol_stateid *stp;
2538
	__be32 ret;
L
Linus Torvalds 已提交
2539 2540 2541 2542

	dprintk("NFSD: nfs4_share_conflict\n");

	fp = find_file(ino);
2543 2544
	if (!fp)
		return nfs_ok;
2545
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2546
	/* Search for conflicting share reservations */
2547
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2548 2549
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2550
			goto out;
L
Linus Torvalds 已提交
2551
	}
2552 2553 2554 2555
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2556 2557
}

2558
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
{
	/* 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);

2569
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2570 2571
	dp->dl_time = get_seconds();

2572 2573 2574
	nfsd4_cb_recall(dp);
}

2575
/* Called from break_lease() with lock_flocks() held. */
2576 2577
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2578 2579
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2580

2581 2582 2583 2584 2585 2586 2587 2588
	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;
	}
2589 2590
	/*
	 * We don't want the locks code to timeout the lease for us;
2591
	 * we'll remove it ourself if a delegation isn't returned
2592
	 * in time:
2593 2594
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2595

2596
	spin_lock(&recall_lock);
2597 2598 2599
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2600
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611
}

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

2612
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2613 2614
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2615 2616
};

2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
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
2627

2628
__be32
A
Andy Adamson 已提交
2629
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2630
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2631 2632 2633 2634
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2635
	struct nfs4_openowner *oo = NULL;
2636
	__be32 status;
L
Linus Torvalds 已提交
2637

2638
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2639
		return nfserr_stale_clientid;
2640 2641 2642 2643 2644 2645 2646
	/*
	 * 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
2647

2648
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2649
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2650 2651
	open->op_openowner = oo;
	if (!oo) {
2652 2653
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2654
		if (clp == NULL)
2655
			return nfserr_expired;
2656
		goto new_owner;
L
Linus Torvalds 已提交
2657
	}
2658
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2659
		/* Replace unconfirmed owners without checking for replay. */
2660 2661 2662
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2663
		goto new_owner;
2664
	}
2665 2666 2667 2668 2669
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2670 2671 2672 2673 2674
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2675 2676 2677 2678
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2679
	return nfs_ok;
L
Linus Torvalds 已提交
2680 2681
}

2682
static inline __be32
N
NeilBrown 已提交
2683 2684 2685 2686 2687 2688 2689 2690
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;
}

2691
static int share_access_to_flags(u32 share_access)
2692
{
2693
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2694 2695
}

2696
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2697
{
2698
	struct nfs4_stid *ret;
2699

2700
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2701 2702 2703
	if (!ret)
		return NULL;
	return delegstateid(ret);
2704 2705
}

2706 2707 2708 2709 2710 2711
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;
}

2712
static __be32
2713
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2714 2715 2716
		struct nfs4_delegation **dp)
{
	int flags;
2717
	__be32 status = nfserr_bad_stateid;
2718

2719
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2720
	if (*dp == NULL)
2721
		goto out;
2722
	flags = share_access_to_flags(open->op_share_access);
2723 2724 2725
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2726
out:
2727
	if (!nfsd4_is_deleg_cur(open))
2728 2729 2730
		return nfs_ok;
	if (status)
		return status;
2731
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2732
	return nfs_ok;
2733 2734
}

2735
static __be32
2736
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2737
{
2738
	struct nfs4_ol_stateid *local;
2739
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2740

2741
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2742 2743 2744 2745
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2746
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2747 2748 2749
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2750
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2751
	}
2752
	return nfs_ok;
L
Linus Torvalds 已提交
2753 2754
}

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

2766 2767
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2768 2769
{
	__be32 status;
2770 2771 2772
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
	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;
}

2784
static inline __be32
L
Linus Torvalds 已提交
2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
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))
2795
		return nfserr_inval;
L
Linus Torvalds 已提交
2796 2797 2798
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2799
static __be32
2800
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
2801
{
2802
	u32 op_share_access = open->op_share_access;
2803
	bool new_access;
2804
	__be32 status;
L
Linus Torvalds 已提交
2805

2806
	new_access = !test_access(op_share_access, stp);
2807
	if (new_access) {
2808
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2809 2810
		if (status)
			return status;
2811
	}
L
Linus Torvalds 已提交
2812 2813
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2814
		if (new_access) {
2815
			int oflag = nfs4_access_to_omode(op_share_access);
2816 2817
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2818 2819 2820
		return status;
	}
	/* remember the open */
2821
	set_access(op_share_access, stp);
2822
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2823 2824 2825 2826 2827 2828

	return nfs_ok;
}


static void
2829
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2830
{
2831
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2832 2833
}

2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846
/* 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;
}

2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858
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;
2859
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2860 2861 2862 2863
	fl->fl_pid = current->tgid;
	return fl;
}

2864 2865
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2866
	struct nfs4_file *fp = dp->dl_file;
2867 2868 2869 2870 2871 2872
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2873
	fl->fl_file = find_readable_file(fp);
2874
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2875
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2876
	if (status) {
2877
		list_del_init(&dp->dl_perclnt);
2878 2879 2880
		locks_free_lock(fl);
		return -ENOMEM;
	}
2881
	fp->fi_lease = fl;
A
Al Viro 已提交
2882
	fp->fi_deleg_file = get_file(fl->fl_file);
2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901
	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);
2902
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2903 2904 2905
	return 0;
}

2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921
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:
2922
			WARN_ON_ONCE(1);
2923 2924 2925 2926
		}
	}
}

L
Linus Torvalds 已提交
2927 2928 2929 2930
/*
 * Attempt to hand out a delegation.
 */
static void
2931 2932
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
2933 2934
{
	struct nfs4_delegation *dp;
2935
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2936
	int cb_up;
2937
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2938

2939
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2940
	flag = NFS4_OPEN_DELEGATE_NONE;
2941 2942 2943
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2944
			if (!cb_up)
2945 2946 2947 2948 2949 2950 2951 2952
				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.... */
2953
			if (locks_in_grace(net))
2954
				goto out;
2955
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2956 2957 2958 2959 2960 2961 2962 2963 2964
				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
2965

2966
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2967 2968
	if (dp == NULL)
		goto out_no_deleg;
2969
	status = nfs4_set_delegation(dp, flag);
2970
	if (status)
2971
		goto out_free;
L
Linus Torvalds 已提交
2972

2973
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2974

2975
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
2976
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2977 2978
out:
	open->op_delegate_type = flag;
2979 2980 2981 2982 2983
	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");

2984 2985 2986
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
2987
	}
2988 2989
	return;
out_free:
2990
	nfs4_put_delegation(dp);
2991 2992 2993
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
2994 2995
}

2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013
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
3014 3015 3016
/*
 * called with nfs4_lock_state() held.
 */
3017
__be32
L
Linus Torvalds 已提交
3018 3019
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3020
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3021
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3022 3023
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3024
	struct nfs4_ol_stateid *stp = NULL;
3025
	struct nfs4_delegation *dp = NULL;
3026
	__be32 status;
L
Linus Torvalds 已提交
3027 3028 3029 3030 3031 3032 3033 3034 3035 3036

	/*
	 * 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;
3037
		status = nfs4_check_deleg(cl, fp, open, &dp);
3038 3039
		if (status)
			goto out;
L
Linus Torvalds 已提交
3040
	} else {
3041
		status = nfserr_bad_stateid;
3042
		if (nfsd4_is_deleg_cur(open))
3043
			goto out;
3044
		status = nfserr_jukebox;
3045 3046 3047
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3048 3049 3050 3051 3052 3053 3054 3055
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3056
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3057 3058 3059
		if (status)
			goto out;
	} else {
3060
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3061 3062 3063
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3064
		if (status)
L
Linus Torvalds 已提交
3065
			goto out;
3066 3067
		stp = open->op_stp;
		open->op_stp = NULL;
3068
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3069
	}
3070 3071
	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
3072

3073
	if (nfsd4_has_session(&resp->cstate)) {
3074
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3075

3076 3077 3078 3079 3080 3081 3082
		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
3083 3084 3085 3086
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3087
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3088
nodeleg:
L
Linus Torvalds 已提交
3089 3090
	status = nfs_ok;

3091
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3092
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3093
out:
3094 3095
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3096 3097
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3098

3099 3100
	if (fp)
		put_nfs4_file(fp);
3101
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3102
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3103 3104 3105 3106
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3107
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3108
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3109 3110 3111 3112 3113
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128
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;
		}
	}
3129 3130
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3131
	if (open->op_stp)
3132
		free_generic_stateid(open->op_stp);
3133 3134
}

3135
__be32
3136 3137
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3138 3139
{
	struct nfs4_client *clp;
3140
	__be32 status;
3141
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3142 3143 3144 3145 3146

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
	status = nfserr_stale_clientid;
3147
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
3148
		goto out;
3149
	clp = find_confirmed_client(clid, cstate->minorversion, nn);
L
Linus Torvalds 已提交
3150 3151 3152 3153 3154 3155 3156
	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;
3157
	if (!list_empty(&clp->cl_delegations)
3158
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3159 3160 3161 3162 3163 3164 3165
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3166
static void
3167
nfsd4_end_grace(struct nfsd_net *nn)
3168
{
3169
	/* do nothing if grace period already ended */
3170
	if (nn->grace_ended)
3171 3172
		return;

3173
	dprintk("NFSD: end of grace period\n");
3174
	nn->grace_ended = true;
3175
	nfsd4_record_grace_done(nn, nn->boot_time);
3176
	locks_end_grace(&nn->nfsd4_manager);
3177 3178 3179 3180 3181
	/*
	 * 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:
	 */
3182
	nn->nfsd4_grace = nn->nfsd4_lease;
3183 3184
}

3185
static time_t
3186
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3187 3188
{
	struct nfs4_client *clp;
3189
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3190 3191
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3192 3193 3194
	time_t cutoff = get_seconds() - nn->nfsd4_lease;
	time_t t, clientid_val = nn->nfsd4_lease;
	time_t u, test_val = nn->nfsd4_lease;
L
Linus Torvalds 已提交
3195 3196 3197 3198

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3199
	nfsd4_end_grace(nn);
3200
	INIT_LIST_HEAD(&reaplist);
3201
	spin_lock(&nn->client_lock);
3202
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3203 3204 3205 3206 3207 3208 3209
		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;
		}
3210 3211 3212 3213 3214 3215 3216
		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);
3217
	}
3218
	spin_unlock(&nn->client_lock);
3219 3220
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3221 3222 3223 3224 3225 3226 3227
		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);
3228 3229
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
		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);
	}
3243
	test_val = nn->nfsd4_lease;
3244
	list_for_each_safe(pos, next, &nn->close_lru) {
3245 3246 3247
		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
3248 3249 3250 3251
			if (test_val > u)
				test_val = u;
			break;
		}
3252
		release_openowner(oo);
L
Linus Torvalds 已提交
3253 3254 3255 3256 3257 3258 3259
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3260 3261 3262 3263
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3264
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3265 3266
{
	time_t t;
3267 3268 3269 3270
	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
3271

3272
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3273
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3274
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3275 3276
}

3277
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3278
{
3279 3280 3281
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3282 3283 3284
}

static int
3285
STALE_STATEID(stateid_t *stateid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3286
{
3287
	if (stateid->si_opaque.so_clid.cl_boot == nn->boot_time)
3288 3289
		return 0;
	dprintk("NFSD: stale stateid " STATEID_FMT "!\n",
3290
		STATEID_VAL(stateid));
3291
	return 1;
L
Linus Torvalds 已提交
3292 3293 3294
}

static inline int
3295
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3296
{
3297 3298 3299
	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
3300 3301 3302
}

static inline int
3303
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3304
{
3305 3306
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3307 3308 3309
}

static
3310
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3311
{
3312
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3313

3314 3315 3316
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3317
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3318
                goto out;
3319
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3320 3321 3322 3323 3324 3325
                goto out;
	status = nfs_ok;
out:
	return status;
}

3326
static inline __be32
3327
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3328
{
3329
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3330
		return nfs_ok;
3331
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3332
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347
		 * 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
3348
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3349
{
3350
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3351 3352
}

3353 3354 3355 3356 3357 3358
/* 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
3359
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3360
{
A
Andy Adamson 已提交
3361 3362 3363 3364
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3365
	if (has_session && in->si_generation == 0)
3366 3367 3368 3369
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3370

3371
	/* If the client sends us a stateid from the future, it's buggy: */
3372
	if (stateid_generation_after(in, ref))
3373 3374
		return nfserr_bad_stateid;
	/*
3375 3376 3377 3378 3379 3380 3381 3382
	 * 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:
3383
	 */
3384
	return nfserr_old_stateid;
3385 3386
}

3387
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3388
{
3389 3390 3391
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3392

3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403
	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;
	}
3404
	s = find_stateid(cl, stateid);
3405
	if (!s)
3406
		return nfserr_bad_stateid;
3407
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3408
	if (status)
3409 3410 3411 3412 3413
		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
3414
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3415 3416
		return nfserr_bad_stateid;
	return nfs_ok;
3417 3418
}

3419 3420 3421
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3422 3423 3424 3425 3426
{
	struct nfs4_client *cl;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
3427
	if (STALE_STATEID(stateid, nn))
3428
		return nfserr_stale_stateid;
3429
	cl = find_confirmed_client(&stateid->si_opaque.so_clid, sessions, nn);
3430 3431 3432 3433 3434 3435 3436 3437 3438
	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
3439 3440 3441
/*
* Checks for stateid operations
*/
3442
__be32
3443
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3444
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3445
{
3446
	struct nfs4_stid *s;
3447
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3448
	struct nfs4_delegation *dp = NULL;
3449
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3450
	struct inode *ino = current_fh->fh_dentry->d_inode;
3451
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3452
	__be32 status;
L
Linus Torvalds 已提交
3453 3454 3455 3456

	if (filpp)
		*filpp = NULL;

3457
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3458 3459 3460
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3461
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3462

3463 3464
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3465 3466
	if (status)
		return status;
3467 3468 3469
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3470 3471
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3472
		dp = delegstateid(s);
3473 3474 3475
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3476
		if (filpp) {
3477
			*filpp = dp->dl_file->fi_deleg_file;
3478 3479 3480 3481 3482
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3483
		}
3484 3485 3486
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3487
		stp = openlockstateid(s);
3488 3489
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3490
			goto out;
3491
		if (stp->st_stateowner->so_is_open_owner
3492
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3493
			goto out;
3494 3495
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3496
			goto out;
3497 3498 3499 3500 3501 3502
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3503 3504 3505
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3506 3507 3508 3509 3510 3511
	}
	status = nfs_ok;
out:
	return status;
}

3512
static __be32
3513
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3514
{
3515
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3516 3517 3518 3519 3520
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3521 3522 3523 3524 3525 3526 3527
/*
 * 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)
{
3528 3529 3530 3531 3532
	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)
3533 3534
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3535 3536
	nfs4_unlock_state();

3537 3538 3539
	return nfs_ok;
}

3540 3541 3542 3543 3544
__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
3545
	struct nfs4_stid *s;
3546
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3547
	__be32 ret = nfserr_bad_stateid;
3548 3549

	nfs4_lock_state();
3550
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3551
	if (!s)
3552
		goto out;
J
J. Bruce Fields 已提交
3553 3554
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3555 3556
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3557 3558 3559 3560 3561 3562 3563 3564 3565
	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;
3566 3567 3568
		break;
	default:
		ret = nfserr_bad_stateid;
3569 3570 3571 3572 3573 3574
	}
out:
	nfs4_unlock_state();
	return ret;
}

3575 3576 3577 3578 3579 3580
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
3581

3582
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3583 3584 3585 3586 3587 3588 3589 3590
{
	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;
3591 3592 3593 3594 3595 3596 3597 3598 3599 3600
	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);
3601 3602
}

L
Linus Torvalds 已提交
3603 3604 3605
/* 
 * Checks for sequence id mutating operations. 
 */
3606
static __be32
3607
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3608
			 stateid_t *stateid, char typemask,
3609 3610
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3611
{
3612
	__be32 status;
3613
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3614

3615 3616
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3617

L
Linus Torvalds 已提交
3618
	*stpp = NULL;
3619 3620
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3621 3622
	if (status)
		return status;
3623
	*stpp = openlockstateid(s);
3624
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3625

3626 3627
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3628

3629 3630
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)
3631 3632 3633
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3634

3635
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3636
						NFS4_OPEN_STID, stpp, nn);
3637 3638
	if (status)
		return status;
3639
	oo = openowner((*stpp)->st_stateowner);
3640
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3641 3642
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3643 3644
}

3645
__be32
3646
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3647
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3648
{
3649
	__be32 status;
3650
	struct nfs4_openowner *oo;
3651
	struct nfs4_ol_stateid *stp;
3652
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3653 3654

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3655 3656
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3657

3658
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3659 3660
	if (status)
		return status;
L
Linus Torvalds 已提交
3661 3662 3663

	nfs4_lock_state();

3664
	status = nfs4_preprocess_seqid_op(cstate,
3665
					oc->oc_seqid, &oc->oc_req_stateid,
3666
					NFS4_OPEN_STID, &stp, nn);
3667
	if (status)
3668
		goto out;
3669
	oo = openowner(stp->st_stateowner);
3670
	status = nfserr_bad_stateid;
3671
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3672
		goto out;
3673
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3674 3675
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3676
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3677
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3678

3679
	nfsd4_client_record_create(oo->oo_owner.so_client);
3680
	status = nfs_ok;
L
Linus Torvalds 已提交
3681
out:
3682 3683
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3684 3685 3686
	return status;
}

J
J. Bruce Fields 已提交
3687
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3688
{
3689
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3690 3691
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3692
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3693
}
3694

J
J. Bruce Fields 已提交
3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708
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:
3709
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3710 3711 3712 3713
	}
}

static void
3714
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3715 3716 3717 3718
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3719
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3720 3721 3722
	}
}

3723
__be32
3724 3725
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3726
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3727
{
3728
	__be32 status;
3729
	struct nfs4_ol_stateid *stp;
3730
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3731 3732

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3733 3734
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3735

3736
	/* We don't yet support WANT bits: */
3737 3738 3739
	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
3740 3741

	nfs4_lock_state();
3742
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3743
					&od->od_stateid, &stp, nn);
3744
	if (status)
L
Linus Torvalds 已提交
3745 3746
		goto out; 
	status = nfserr_inval;
3747 3748
	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
3749 3750 3751
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3752
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3753 3754 3755 3756
		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
3757
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3758

3759
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3760

3761 3762
	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
3763 3764
	status = nfs_ok;
out:
3765 3766
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3767 3768 3769
	return status;
}

3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786
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;
3787 3788 3789 3790 3791 3792 3793
	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;
3794 3795
}

L
Linus Torvalds 已提交
3796 3797 3798
/*
 * nfs4_unlock_state() called after encode
 */
3799
__be32
3800
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3801
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3802
{
3803
	__be32 status;
3804
	struct nfs4_openowner *oo;
3805
	struct nfs4_ol_stateid *stp;
3806 3807
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3808 3809

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3810 3811
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3812 3813

	nfs4_lock_state();
3814 3815 3816
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3817
					&stp, nn);
3818
	if (status)
L
Linus Torvalds 已提交
3819
		goto out; 
3820
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3821
	status = nfs_ok;
3822 3823
	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
3824

3825
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3826
	release_last_closed_stateid(oo);
3827
	oo->oo_last_closed_stid = stp;
3828

3829 3830 3831 3832 3833 3834 3835 3836 3837 3838
	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))
3839
				move_to_close_lru(oo, SVC_NET(rqstp));
3840 3841
		}
	}
L
Linus Torvalds 已提交
3842
out:
3843 3844
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3845 3846 3847
	return status;
}

3848
__be32
3849 3850
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3851
{
3852 3853
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3854
	struct nfs4_stid *s;
3855
	__be32 status;
3856
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3857

3858
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3859
		return status;
L
Linus Torvalds 已提交
3860 3861

	nfs4_lock_state();
3862 3863
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3864
	if (status)
3865
		goto out;
3866
	dp = delegstateid(s);
3867
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3868 3869 3870 3871
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3872
out:
3873 3874
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3875 3876 3877 3878 3879 3880
	return status;
}


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

3881
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3882

B
Benny Halevy 已提交
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897
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;

3898
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3899 3900 3901 3902
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3903
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3904 3905 3906
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3907
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926
}

/*
 * 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;
}

3927 3928
/* 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. */
3929
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3930
};
L
Linus Torvalds 已提交
3931 3932 3933 3934

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3935
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3936

3937
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3938 3939 3940
		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);
3941 3942 3943
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3944 3945
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3946
	} else {
3947 3948 3949
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3950 3951
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3952 3953
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3954 3955
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3956 3957 3958 3959 3960 3961
		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
3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972
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;
}

3973 3974
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3975
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3976
{
3977
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
3978
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3979

3980
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
3981 3982
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
3983 3984 3985 3986
	}
	return NULL;
}

3987
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
3988
{
3989 3990 3991
	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);
3992
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
3993

3994
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
3995
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
3996
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
3997 3998
}

L
Linus Torvalds 已提交
3999 4000 4001
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4002
 * occurred. 
L
Linus Torvalds 已提交
4003
 *
4004
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4005 4006
 */

4007
static struct nfs4_lockowner *
4008
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4009
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4010

4011 4012
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4013
		return NULL;
4014 4015
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4016 4017
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4018 4019 4020
	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
4021 4022
}

4023 4024
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
4025
{
4026
	struct nfs4_ol_stateid *stp;
4027
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4028

4029
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4030
	if (stp == NULL)
J
J. Bruce Fields 已提交
4031
		return NULL;
4032
	init_stid(&stp->st_stid, clp, NFS4_LOCK_STID);
4033
	list_add(&stp->st_perfile, &fp->fi_stateids);
4034 4035
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4036
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4037
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4038
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4039
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4040
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4041 4042 4043
	return stp;
}

4044
static int
L
Linus Torvalds 已提交
4045 4046
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4047
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4048 4049 4050
	     LOFF_OVERFLOW(offset, length)));
}

4051
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4052 4053 4054 4055
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4056
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4057 4058
		return;
	nfs4_file_get_access(fp, oflag);
4059
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4060 4061
}

J
J. Bruce Fields 已提交
4062
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)
4063 4064 4065 4066 4067 4068
{
	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;
4069
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4070

4071 4072
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4073 4074 4075 4076 4077 4078 4079 4080
	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;
	}
4081
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094
			&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
4095 4096 4097
/*
 *  LOCK operation 
 */
4098
__be32
4099
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4100
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4101
{
4102 4103
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4104
	struct nfs4_ol_stateid *lock_stp;
4105
	struct file *filp = NULL;
4106 4107
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4108
	__be32 status = 0;
4109
	bool new_state = false;
4110
	int lkflg;
4111
	int err;
4112 4113
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4114 4115 4116 4117 4118 4119 4120 4121

	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;

4122
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4123
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4124 4125 4126 4127
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4128 4129 4130
	nfs4_lock_state();

	if (lock->lk_is_new) {
4131
		struct nfs4_ol_stateid *open_stp = NULL;
4132 4133 4134 4135 4136 4137 4138

		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
4139
		status = nfserr_stale_clientid;
4140
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4141 4142 4143
			goto out;

		/* validate and update open stateid and open seqid */
4144
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4145 4146
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4147
					&open_stp, nn);
4148
		if (status)
L
Linus Torvalds 已提交
4149
			goto out;
4150
		open_sop = openowner(open_stp->st_stateowner);
4151
		status = nfserr_bad_stateid;
4152
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4153 4154
						&lock->v.new.clientid))
			goto out;
4155 4156
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4157
	} else
4158
		status = nfs4_preprocess_seqid_op(cstate,
4159 4160
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4161
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4162 4163
	if (status)
		goto out;
4164
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4165

4166 4167 4168 4169 4170
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4171
	status = nfserr_grace;
4172
	if (locks_in_grace(net) && !lock->lk_reclaim)
4173 4174
		goto out;
	status = nfserr_no_grace;
4175
	if (!locks_in_grace(net) && lock->lk_reclaim)
4176 4177
		goto out;

4178 4179 4180 4181 4182 4183 4184 4185
	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
4186 4187 4188
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4189 4190 4191
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4192
			file_lock->fl_type = F_RDLCK;
4193
			break;
L
Linus Torvalds 已提交
4194 4195
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4196 4197 4198
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4199
			file_lock->fl_type = F_WRLCK;
4200
			break;
L
Linus Torvalds 已提交
4201 4202 4203 4204
		default:
			status = nfserr_inval;
		goto out;
	}
4205 4206 4207 4208
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223
	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
4224

4225
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4226
	switch (-err) {
L
Linus Torvalds 已提交
4227
	case 0: /* success! */
4228 4229
		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
4230
				sizeof(stateid_t));
4231
		status = 0;
4232 4233 4234 4235
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4236
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4237
		break;
L
Linus Torvalds 已提交
4238 4239
	case (EDEADLK):
		status = nfserr_deadlock;
4240
		break;
4241
	default:
4242
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4243
		status = nfserrno(err);
4244
		break;
L
Linus Torvalds 已提交
4245 4246
	}
out:
4247
	if (status && new_state)
4248
		release_lockowner(lock_sop);
4249 4250
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4251 4252 4253 4254
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4255 4256 4257
	return status;
}

4258 4259 4260 4261 4262 4263
/*
 * 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.)
 */
4264
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4265 4266
{
	struct file *file;
4267 4268 4269 4270 4271
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4272 4273 4274
	return err;
}

L
Linus Torvalds 已提交
4275 4276 4277
/*
 * LOCKT operation
 */
4278
__be32
4279 4280
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4281 4282
{
	struct inode *inode;
4283
	struct file_lock *file_lock = NULL;
4284
	struct nfs4_lockowner *lo;
4285
	__be32 status;
4286
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4287

4288
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4289 4290 4291 4292 4293 4294 4295 4296
		return nfserr_grace;

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

	nfs4_lock_state();

	status = nfserr_stale_clientid;
4297
	if (!nfsd4_has_session(cstate) && STALE_CLIENTID(&lockt->lt_clientid, nn))
L
Linus Torvalds 已提交
4298 4299
		goto out;

4300
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4301 4302
		goto out;

4303
	inode = cstate->current_fh.fh_dentry->d_inode;
4304 4305 4306 4307 4308 4309 4310
	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
4311 4312 4313
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4314
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4315 4316 4317
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4318
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4319 4320
		break;
		default:
4321
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4322 4323 4324 4325
			status = nfserr_inval;
		goto out;
	}

4326
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4327
	if (lo)
4328 4329 4330
		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
4331

4332 4333
	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
4334

4335
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4336

4337
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4338
	if (status)
4339
		goto out;
4340

4341
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4342
		status = nfserr_denied;
4343
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4344 4345 4346
	}
out:
	nfs4_unlock_state();
4347 4348
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4349 4350 4351
	return status;
}

4352
__be32
4353
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4354
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4355
{
4356
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4357
	struct file *filp = NULL;
4358
	struct file_lock *file_lock = NULL;
4359
	__be32 status;
4360
	int err;
4361 4362
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4363 4364 4365 4366 4367 4368 4369 4370 4371
	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();
									        
4372
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4373 4374
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4375
	if (status)
L
Linus Torvalds 已提交
4376
		goto out;
4377 4378 4379 4380 4381
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399
	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
4400 4401 4402 4403

	/*
	*  Try to unlock the file in the VFS.
	*/
4404
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4405
	if (err) {
4406
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4407 4408 4409 4410 4411
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4412 4413
	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
4414 4415

out:
4416 4417
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4418 4419
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4420 4421 4422
	return status;

out_nfserr:
4423
	status = nfserrno(err);
L
Linus Torvalds 已提交
4424 4425 4426 4427 4428 4429 4430 4431 4432
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4433
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4434 4435
{
	struct file_lock **flpp;
4436
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4437 4438
	int status = 0;

4439
	lock_flocks();
L
Linus Torvalds 已提交
4440
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4441
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4442 4443
			status = 1;
			goto out;
4444
		}
L
Linus Torvalds 已提交
4445 4446
	}
out:
4447
	unlock_flocks();
L
Linus Torvalds 已提交
4448 4449 4450
	return status;
}

4451
__be32
4452 4453 4454
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4455 4456
{
	clientid_t *clid = &rlockowner->rl_clientid;
4457
	struct nfs4_stateowner *sop;
4458
	struct nfs4_lockowner *lo;
4459
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4460
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4461
	struct list_head matches;
4462
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4463
	__be32 status;
4464
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4465 4466 4467 4468 4469 4470 4471

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

	/* XXX check for lease expiration */

	status = nfserr_stale_clientid;
4472
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
4473 4474 4475 4476
		return status;

	nfs4_lock_state();

4477 4478
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4479

4480
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4481 4482
		if (sop->so_is_open_owner)
			continue;
4483 4484 4485 4486 4487 4488 4489 4490
		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
4491
		}
4492 4493 4494 4495 4496
	}
	/* 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
4497
	while (!list_empty(&matches)) {
4498 4499
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4500 4501
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4502 4503
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4504 4505 4506 4507 4508 4509 4510
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4511
alloc_reclaim(void)
L
Linus Torvalds 已提交
4512
{
N
NeilBrown 已提交
4513
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4514 4515
}

4516
bool
4517
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4518
{
4519
	struct nfs4_client_reclaim *crp;
4520

4521
	crp = nfsd4_find_reclaim_client(name, nn);
4522
	return (crp && crp->cr_clp);
4523 4524
}

L
Linus Torvalds 已提交
4525 4526 4527
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4528
struct nfs4_client_reclaim *
4529
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4530 4531
{
	unsigned int strhashval;
4532
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4533

N
NeilBrown 已提交
4534 4535
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4536 4537 4538
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4539
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4540
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4541
		crp->cr_clp = NULL;
4542
		nn->reclaim_str_hashtbl_size++;
4543 4544
	}
	return crp;
L
Linus Torvalds 已提交
4545 4546
}

4547
void
4548
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4549 4550 4551
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4552
	nn->reclaim_str_hashtbl_size--;
4553 4554
}

4555
void
4556
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4557 4558 4559 4560 4561
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4562 4563
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4564
			                struct nfs4_client_reclaim, cr_strhash);
4565
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4566 4567
		}
	}
4568
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4569 4570 4571 4572
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4573
struct nfs4_client_reclaim *
4574
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4575 4576 4577 4578
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4579
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4580

4581
	strhashval = clientstr_hashval(recdir);
4582
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4583
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4584 4585 4586 4587 4588 4589 4590 4591 4592
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4593
__be32
4594
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4595
{
4596 4597 4598
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4599
	clp = find_confirmed_client(clid, sessions, nn);
4600 4601 4602 4603
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4604 4605
}

B
Bryan Schumaker 已提交
4606 4607
#ifdef CONFIG_NFSD_FAULT_INJECTION

4608 4609 4610 4611 4612 4613
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
	expire_client(clp);
	return 1;
}

B
Bryan Schumaker 已提交
4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639
static u64 nfsd_foreach_client_lock(struct nfs4_client *clp, u64 max, void (*func)(struct nfs4_lockowner *))
{
	struct nfs4_openowner *oop;
	struct nfs4_lockowner *lop, *lo_next;
	struct nfs4_ol_stateid *stp, *st_next;
	u64 count = 0;

	list_for_each_entry(oop, &clp->cl_openowners, oo_perclient) {
		list_for_each_entry_safe(stp, st_next, &oop->oo_owner.so_stateids, st_perstateowner) {
			list_for_each_entry_safe(lop, lo_next, &stp->st_lockowners, lo_perstateid) {
				if (func)
					func(lop);
				if (++count == max)
					return count;
			}
		}
	}

	return count;
}

u64 nfsd_forget_client_locks(struct nfs4_client *clp, u64 max)
{
	return nfsd_foreach_client_lock(clp, max, release_lockowner);
}

4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659
static u64 nfsd_foreach_client_open(struct nfs4_client *clp, u64 max, void (*func)(struct nfs4_openowner *))
{
	struct nfs4_openowner *oop, *next;
	u64 count = 0;

	list_for_each_entry_safe(oop, next, &clp->cl_openowners, oo_perclient) {
		if (func)
			func(oop);
		if (++count == max)
			break;
	}

	return count;
}

u64 nfsd_forget_client_openowners(struct nfs4_client *clp, u64 max)
{
	return nfsd_foreach_client_open(clp, max, release_openowner);
}

4660
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4661 4662
{
	struct nfs4_client *clp, *next;
4663
	u64 count = 0;
4664
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4665

4666 4667 4668
	if (!nfsd_netns_ready(nn))
		return 0;

4669
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4670 4671
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4672 4673 4674
			break;
	}

4675 4676 4677 4678 4679 4680 4681
	return count;
}

void nfsd_forget_clients(u64 num)
{
	u64 count = nfsd_for_n_state(num, nfsd_forget_client);
	printk(KERN_INFO "NFSD: Forgot %llu clients", count);
B
Bryan Schumaker 已提交
4682 4683 4684 4685
}

void nfsd_forget_locks(u64 num)
{
B
Bryan Schumaker 已提交
4686 4687
	u64 count = nfsd_for_n_state(num, nfsd_forget_client_locks);
	printk(KERN_INFO "NFSD: Forgot %llu locks", count);
B
Bryan Schumaker 已提交
4688 4689 4690 4691
}

void nfsd_forget_openowners(u64 num)
{
4692 4693
	u64 count = nfsd_for_n_state(num, nfsd_forget_client_openowners);
	printk(KERN_INFO "NFSD: Forgot %llu open owners", count);
B
Bryan Schumaker 已提交
4694 4695
}

4696
static int nfsd_process_n_delegations(u64 num, struct list_head *list)
B
Bryan Schumaker 已提交
4697 4698
{
	int i, count = 0;
4699 4700
	struct nfs4_file *fp, *fnext;
	struct nfs4_delegation *dp, *dnext;
B
Bryan Schumaker 已提交
4701 4702

	for (i = 0; i < FILE_HASH_SIZE; i++) {
4703 4704
		list_for_each_entry_safe(fp, fnext, &file_hashtbl[i], fi_hash) {
			list_for_each_entry_safe(dp, dnext, &fp->fi_delegations, dl_perfile) {
4705
				list_move(&dp->dl_recall_lru, list);
B
Bryan Schumaker 已提交
4706 4707 4708 4709 4710
				if (++count == num)
					return count;
			}
		}
	}
4711

B
Bryan Schumaker 已提交
4712 4713 4714 4715 4716 4717
	return count;
}

void nfsd_forget_delegations(u64 num)
{
	unsigned int count;
4718 4719 4720 4721 4722 4723
	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
4724

4725 4726
	list_for_each_entry_safe(dp, dnext, &victims, dl_recall_lru)
		unhash_delegation(dp);
B
Bryan Schumaker 已提交
4727 4728 4729 4730 4731 4732 4733

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

void nfsd_recall_delegations(u64 num)
{
	unsigned int count;
4734 4735
	LIST_HEAD(victims);
	struct nfs4_delegation *dp, *dnext;
B
Bryan Schumaker 已提交
4736 4737

	spin_lock(&recall_lock);
4738 4739 4740 4741 4742
	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
4743 4744 4745 4746 4747 4748 4749
	spin_unlock(&recall_lock);

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

#endif /* CONFIG_NFSD_FAULT_INJECTION */

4750
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4751

4752
void
4753
nfs4_state_init(void)
L
Linus Torvalds 已提交
4754
{
4755
	int i;
L
Linus Torvalds 已提交
4756 4757 4758 4759 4760

	for (i = 0; i < FILE_HASH_SIZE; i++) {
		INIT_LIST_HEAD(&file_hashtbl[i]);
	}
	INIT_LIST_HEAD(&del_recall_lru);
4761 4762
}

4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783
/*
 * 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);
}

4784
static int nfs4_state_create_net(struct net *net)
4785 4786 4787 4788 4789 4790 4791
{
	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)
4792
		goto err;
4793 4794 4795 4796
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4797 4798 4799 4800
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4801 4802 4803 4804
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4805 4806 4807 4808
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4809

4810
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4811
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4812
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4813
	}
4814 4815
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4816 4817
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4818 4819
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4820
	nn->conf_name_tree = RB_ROOT;
4821
	nn->unconf_name_tree = RB_ROOT;
4822
	INIT_LIST_HEAD(&nn->client_lru);
4823
	INIT_LIST_HEAD(&nn->close_lru);
4824
	spin_lock_init(&nn->client_lock);
4825

4826
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4827
	get_net(net);
4828

4829
	return 0;
4830

4831 4832
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4833 4834
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4835 4836
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4837 4838
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4839 4840
err:
	return -ENOMEM;
4841 4842 4843
}

static void
4844
nfs4_state_destroy_net(struct net *net)
4845 4846 4847 4848
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4849
	struct rb_node *node, *tmp;
4850 4851 4852 4853 4854 4855 4856

	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);
		}
	}
4857 4858 4859 4860 4861 4862 4863 4864 4865 4866

	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);
	}

4867
	kfree(nn->sessionid_hashtbl);
4868
	kfree(nn->lockowner_ino_hashtbl);
4869
	kfree(nn->ownerstr_hashtbl);
4870
	kfree(nn->unconf_id_hashtbl);
4871
	kfree(nn->conf_id_hashtbl);
4872
	put_net(net);
4873 4874
}

4875
int
4876
nfs4_state_start_net(struct net *net)
4877
{
4878
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4879 4880
	int ret;

4881 4882 4883 4884 4885 4886 4887 4888 4889 4890
	/*
	 * 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.
	 */
	if (net != &init_net)
		return -EINVAL;

4891
	ret = nfs4_state_create_net(net);
4892 4893
	if (ret)
		return ret;
4894
	nfsd4_client_tracking_init(net);
4895
	nn->boot_time = get_seconds();
4896
	locks_start_grace(net, &nn->nfsd4_manager);
4897
	nn->grace_ended = false;
4898
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4899 4900
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4901 4902 4903 4904 4905 4906 4907 4908 4909 4910
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

4911
	ret = set_callback_cred();
4912 4913
	if (ret)
		return -ENOMEM;
4914
	laundry_wq = create_singlethread_workqueue("nfsd4");
4915 4916 4917 4918
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4919 4920 4921
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4922

4923
	set_max_delegations();
4924

4925
	return 0;
4926

4927 4928
out_free_laundry:
	destroy_workqueue(laundry_wq);
4929
out_recovery:
4930
	return ret;
L
Linus Torvalds 已提交
4931 4932
}

4933
/* should be called with the state lock held */
4934
void
4935
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4936 4937 4938
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4939
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4940

4941 4942
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4943

L
Linus Torvalds 已提交
4944 4945 4946 4947
	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);
4948 4949
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4950 4951 4952 4953 4954 4955 4956 4957
		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);
	}

4958
	nfsd4_client_tracking_exit(net);
4959
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
4960 4961 4962 4963 4964
}

void
nfs4_state_shutdown(void)
{
4965
	destroy_workqueue(laundry_wq);
4966
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
4967
}
4968 4969 4970 4971

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4972 4973
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
4974 4975 4976 4977 4978
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
4979 4980 4981 4982 4983 4984 4985 4986 4987 4988
	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);
4989 4990
}

4991 4992 4993
/*
 * functions to set current state id
 */
4994 4995 4996 4997 4998 4999
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5000 5001 5002 5003 5004 5005
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020
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
 */
5021

5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033
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);
}

5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045
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);
}

5046 5047 5048 5049 5050 5051 5052
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5053
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5054
{
5055
	get_stateid(cstate, &locku->lu_stateid);
5056
}
5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068

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);
}