nfs4state.c 133.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
/*
*  Copyright (c) 2001 The Regents of the University of Michigan.
*  All rights reserved.
*
*  Kendrick Smith <kmsmith@umich.edu>
*  Andy Adamson <kandros@umich.edu>
*
*  Redistribution and use in source and binary forms, with or without
*  modification, are permitted provided that the following conditions
*  are met:
*
*  1. Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*  2. Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in the
*     documentation and/or other materials provided with the distribution.
*  3. Neither the name of the University nor the names of its
*     contributors may be used to endorse or promote products derived
*     from this software without specific prior written permission.
*
*  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
*  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
*  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
*  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
*  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
*  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
*  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
*  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
*  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
*  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
*  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/

35
#include <linux/file.h>
36
#include <linux/fs.h>
37
#include <linux/slab.h>
38
#include <linux/namei.h>
39
#include <linux/swap.h>
40
#include <linux/pagemap.h>
41
#include <linux/ratelimit.h>
42
#include <linux/sunrpc/svcauth_gss.h>
43
#include <linux/sunrpc/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
}

746 747 748 749
static void init_forechannel_attrs(struct nfsd4_channel_attrs *new,
				   struct nfsd4_channel_attrs *req,
				   int numslots, int slotsize,
				   struct nfsd_net *nn)
A
Andy Adamson 已提交
750
{
751
	u32 maxrpc = nn->nfsd_serv->sv_max_mesg;
A
Andy Adamson 已提交
752

753
	new->maxreqs = numslots;
754 755
	new->maxresp_cached = min_t(u32, req->maxresp_cached,
					slotsize + NFSD_MIN_HDR_SEQ_SZ);
756 757 758 759
	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
760

761 762 763 764 765
static void free_conn(struct nfsd4_conn *c)
{
	svc_xprt_put(c->cn_xprt);
	kfree(c);
}
A
Andy Adamson 已提交
766

767 768 769 770
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
771

772 773 774 775 776 777
	spin_lock(&clp->cl_lock);
	if (!list_empty(&c->cn_persession)) {
		list_del(&c->cn_persession);
		free_conn(c);
	}
	spin_unlock(&clp->cl_lock);
778
	nfsd4_probe_callback(clp);
779
}
A
Andy Adamson 已提交
780

781
static struct nfsd4_conn *alloc_conn(struct svc_rqst *rqstp, u32 flags)
782 783
{
	struct nfsd4_conn *conn;
A
Andy Adamson 已提交
784

785 786
	conn = kmalloc(sizeof(struct nfsd4_conn), GFP_KERNEL);
	if (!conn)
787
		return NULL;
788 789
	svc_xprt_get(rqstp->rq_xprt);
	conn->cn_xprt = rqstp->rq_xprt;
790
	conn->cn_flags = flags;
791 792 793
	INIT_LIST_HEAD(&conn->cn_xpt_user.list);
	return conn;
}
794

795 796 797 798
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
799 800
}

801
static void nfsd4_hash_conn(struct nfsd4_conn *conn, struct nfsd4_session *ses)
802
{
803
	struct nfs4_client *clp = ses->se_client;
804

805
	spin_lock(&clp->cl_lock);
806
	__nfsd4_hash_conn(conn, ses);
807
	spin_unlock(&clp->cl_lock);
808 809
}

810
static int nfsd4_register_conn(struct nfsd4_conn *conn)
811
{
812
	conn->cn_xpt_user.callback = nfsd4_conn_lost;
813
	return register_xpt_user(conn->cn_xprt, &conn->cn_xpt_user);
814 815
}

816
static void nfsd4_init_conn(struct svc_rqst *rqstp, struct nfsd4_conn *conn, struct nfsd4_session *ses)
A
Andy Adamson 已提交
817
{
818
	int ret;
A
Andy Adamson 已提交
819

820
	nfsd4_hash_conn(conn, ses);
821 822 823 824
	ret = nfsd4_register_conn(conn);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&conn->cn_xpt_user);
825
	if (conn->cn_flags & NFS4_CDFC4_BACK) {
826 827 828
		/* callback channel may be back up */
		nfsd4_probe_callback(ses->se_client);
	}
829
}
A
Andy Adamson 已提交
830

831
static struct nfsd4_conn *alloc_conn_from_crses(struct svc_rqst *rqstp, struct nfsd4_create_session *cses)
832 833 834
{
	u32 dir = NFS4_CDFC4_FORE;

835
	if (cses->flags & SESSION4_BACK_CHAN)
836
		dir |= NFS4_CDFC4_BACK;
837
	return alloc_conn(rqstp, dir);
838 839 840
}

/* must be called under client_lock */
841
static void nfsd4_del_conns(struct nfsd4_session *s)
842
{
843 844
	struct nfs4_client *clp = s->se_client;
	struct nfsd4_conn *c;
A
Andy Adamson 已提交
845

846 847 848 849 850
	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);
851

852 853
		unregister_xpt_user(c->cn_xprt, &c->cn_xpt_user);
		free_conn(c);
A
Andy Adamson 已提交
854

855 856 857
		spin_lock(&clp->cl_lock);
	}
	spin_unlock(&clp->cl_lock);
858
}
A
Andy Adamson 已提交
859

860 861 862 863 864 865 866
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);
}

867
static void free_session(struct kref *kref)
868 869
{
	struct nfsd4_session *ses;
870
	struct nfsd_net *nn;
871 872

	ses = container_of(kref, struct nfsd4_session, se_ref);
873 874 875
	nn = net_generic(ses->se_client->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
876
	nfsd4_del_conns(ses);
877
	__free_session(ses);
878 879
}

880 881
void nfsd4_put_session(struct nfsd4_session *ses)
{
882 883 884
	struct nfsd_net *nn = net_generic(ses->se_client->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
885
	nfsd4_put_session_locked(ses);
886
	spin_unlock(&nn->client_lock);
887 888
}

889 890
static struct nfsd4_session *alloc_session(struct nfsd4_channel_attrs *fchan,
					   struct nfsd_net *nn)
891 892 893 894 895 896 897 898 899 900 901 902
{
	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);
903 904
	if (numslots < 1)
		return NULL;
905

906
	new = __alloc_session(slotsize, numslots);
907 908
	if (!new) {
		nfsd4_put_drc_mem(slotsize, fchan->maxreqs);
909
		return NULL;
910
	}
911
	init_forechannel_attrs(&new->se_fchannel, fchan, numslots, slotsize, nn);
912 913
	return new;
}
914

915
static void init_session(struct svc_rqst *rqstp, struct nfsd4_session *new, struct nfs4_client *clp, struct nfsd4_create_session *cses)
916 917
{
	int idx;
918
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
919

A
Andy Adamson 已提交
920 921 922
	new->se_client = clp;
	gen_sessionid(new);

923 924
	INIT_LIST_HEAD(&new->se_conns);

925
	new->se_cb_seq_nr = 1;
A
Andy Adamson 已提交
926
	new->se_flags = cses->flags;
927
	new->se_cb_prog = cses->callback_prog;
928
	new->se_cb_sec = cses->cb_sec;
A
Andy Adamson 已提交
929
	kref_init(&new->se_ref);
930
	idx = hash_sessionid(&new->se_sessionid);
931
	spin_lock(&nn->client_lock);
932
	list_add(&new->se_hash, &nn->sessionid_hashtbl[idx]);
933
	spin_lock(&clp->cl_lock);
A
Andy Adamson 已提交
934
	list_add(&new->se_perclnt, &clp->cl_sessions);
935
	spin_unlock(&clp->cl_lock);
936
	spin_unlock(&nn->client_lock);
A
Andy Adamson 已提交
937

938
	if (cses->flags & SESSION4_BACK_CHAN) {
939
		struct sockaddr *sa = svc_addr(rqstp);
940 941 942 943 944 945 946
		/*
		 * 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:
		 */
947 948 949
		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
950 951
}

952
/* caller must hold client_lock */
M
Marc Eshel 已提交
953
static struct nfsd4_session *
954
find_in_sessionid_hashtbl(struct nfs4_sessionid *sessionid, struct net *net)
M
Marc Eshel 已提交
955 956 957
{
	struct nfsd4_session *elem;
	int idx;
958
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
M
Marc Eshel 已提交
959 960 961 962

	dump_sessionid(__func__, sessionid);
	idx = hash_sessionid(sessionid);
	/* Search in the appropriate list */
963
	list_for_each_entry(elem, &nn->sessionid_hashtbl[idx], se_hash) {
M
Marc Eshel 已提交
964 965 966 967 968 969 970 971 972 973
		if (!memcmp(elem->se_sessionid.data, sessionid->data,
			    NFS4_MAX_SESSIONID_LEN)) {
			return elem;
		}
	}

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

974
/* caller must hold client_lock */
A
Andy Adamson 已提交
975
static void
M
Marc Eshel 已提交
976
unhash_session(struct nfsd4_session *ses)
A
Andy Adamson 已提交
977 978
{
	list_del(&ses->se_hash);
979
	spin_lock(&ses->se_client->cl_lock);
A
Andy Adamson 已提交
980
	list_del(&ses->se_perclnt);
981
	spin_unlock(&ses->se_client->cl_lock);
M
Marc Eshel 已提交
982 983
}

984
/* must be called under the client_lock */
L
Linus Torvalds 已提交
985
static inline void
986
renew_client_locked(struct nfs4_client *clp)
L
Linus Torvalds 已提交
987
{
988 989
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

B
Benny Halevy 已提交
990
	if (is_client_expired(clp)) {
991 992
		WARN_ON(1);
		printk("%s: client (clientid %08x/%08x) already expired\n",
B
Benny Halevy 已提交
993 994 995 996 997 998
			__func__,
			clp->cl_clientid.cl_boot,
			clp->cl_clientid.cl_id);
		return;
	}

L
Linus Torvalds 已提交
999 1000 1001
	dprintk("renewing client (clientid %08x/%08x)\n", 
			clp->cl_clientid.cl_boot, 
			clp->cl_clientid.cl_id);
1002
	list_move_tail(&clp->cl_lru, &nn->client_lru);
L
Linus Torvalds 已提交
1003 1004 1005
	clp->cl_time = get_seconds();
}

1006 1007 1008
static inline void
renew_client(struct nfs4_client *clp)
{
1009 1010 1011
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	spin_lock(&nn->client_lock);
1012
	renew_client_locked(clp);
1013
	spin_unlock(&nn->client_lock);
1014 1015
}

L
Linus Torvalds 已提交
1016 1017
/* SETCLIENTID and SETCLIENTID_CONFIRM Helper functions */
static int
1018
STALE_CLIENTID(clientid_t *clid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1019
{
1020
	if (clid->cl_boot == nn->boot_time)
L
Linus Torvalds 已提交
1021
		return 0;
A
Andy Adamson 已提交
1022
	dprintk("NFSD stale clientid (%08x/%08x) boot_time %08lx\n",
1023
		clid->cl_boot, clid->cl_id, nn->boot_time);
L
Linus Torvalds 已提交
1024 1025 1026 1027 1028 1029 1030 1031
	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.
 */
1032
static struct nfs4_client *alloc_client(struct xdr_netobj name)
L
Linus Torvalds 已提交
1033 1034 1035
{
	struct nfs4_client *clp;

1036 1037 1038
	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
	if (clp == NULL)
		return NULL;
1039
	clp->cl_name.data = kmemdup(name.data, name.len, GFP_KERNEL);
1040 1041 1042
	if (clp->cl_name.data == NULL) {
		kfree(clp);
		return NULL;
L
Linus Torvalds 已提交
1043
	}
1044
	clp->cl_name.len = name.len;
L
Linus Torvalds 已提交
1045 1046 1047 1048 1049 1050
	return clp;
}

static inline void
free_client(struct nfs4_client *clp)
{
1051 1052 1053
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	lockdep_assert_held(&nn->client_lock);
1054 1055 1056 1057 1058
	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);
1059
		nfsd4_put_session_locked(ses);
1060
	}
1061
	free_svc_cred(&clp->cl_cred);
L
Linus Torvalds 已提交
1062 1063 1064 1065
	kfree(clp->cl_name.data);
	kfree(clp);
}

1066 1067 1068 1069
void
release_session_client(struct nfsd4_session *session)
{
	struct nfs4_client *clp = session->se_client;
1070
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
1071

1072
	if (!atomic_dec_and_lock(&clp->cl_refcount, &nn->client_lock))
1073 1074 1075 1076 1077 1078
		return;
	if (is_client_expired(clp)) {
		free_client(clp);
		session->se_client = NULL;
	} else
		renew_client_locked(clp);
1079
	spin_unlock(&nn->client_lock);
1080 1081
}

B
Benny Halevy 已提交
1082 1083 1084 1085
/* must be called under the client_lock */
static inline void
unhash_client_locked(struct nfs4_client *clp)
{
1086 1087
	struct nfsd4_session *ses;

B
Benny Halevy 已提交
1088
	mark_client_expired(clp);
B
Benny Halevy 已提交
1089
	list_del(&clp->cl_lru);
1090
	spin_lock(&clp->cl_lock);
1091 1092
	list_for_each_entry(ses, &clp->cl_sessions, se_perclnt)
		list_del_init(&ses->se_hash);
1093
	spin_unlock(&clp->cl_lock);
B
Benny Halevy 已提交
1094 1095
}

L
Linus Torvalds 已提交
1096
static void
1097
destroy_client(struct nfs4_client *clp)
L
Linus Torvalds 已提交
1098
{
1099
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
1100 1101
	struct nfs4_delegation *dp;
	struct list_head reaplist;
1102
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1103 1104 1105

	INIT_LIST_HEAD(&reaplist);
	spin_lock(&recall_lock);
1106 1107 1108
	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
1109 1110 1111 1112 1113 1114 1115
		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);
	}
1116
	while (!list_empty(&clp->cl_openowners)) {
1117 1118
		oo = list_entry(clp->cl_openowners.next, struct nfs4_openowner, oo_perclient);
		release_openowner(oo);
L
Linus Torvalds 已提交
1119
	}
1120
	nfsd4_shutdown_callback(clp);
B
Benny Halevy 已提交
1121 1122
	if (clp->cl_cb_conn.cb_xprt)
		svc_xprt_put(clp->cl_cb_conn.cb_xprt);
1123
	list_del(&clp->cl_idhash);
1124
	if (test_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags))
1125
		rb_erase(&clp->cl_namenode, &nn->conf_name_tree);
1126
	else
1127
		rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1128
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
1129
	unhash_client_locked(clp);
1130 1131
	if (atomic_read(&clp->cl_refcount) == 0)
		free_client(clp);
1132
	spin_unlock(&nn->client_lock);
L
Linus Torvalds 已提交
1133 1134
}

1135 1136 1137 1138 1139 1140
static void expire_client(struct nfs4_client *clp)
{
	nfsd4_client_record_remove(clp);
	destroy_client(clp);
}

1141 1142 1143 1144
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
1145 1146
}

1147 1148
static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
{
L
Linus Torvalds 已提交
1149 1150 1151 1152
	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
}

1153
static int copy_cred(struct svc_cred *target, struct svc_cred *source)
1154
{
1155 1156 1157 1158 1159 1160 1161
	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;
1162
	target->cr_flavor = source->cr_flavor;
L
Linus Torvalds 已提交
1163 1164 1165 1166
	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);
1167
	return 0;
L
Linus Torvalds 已提交
1168 1169
}

1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
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);
}

1181
static int same_name(const char *n1, const char *n2)
1182
{
N
NeilBrown 已提交
1183
	return 0 == memcmp(n1, n2, HEXDIR_LEN);
L
Linus Torvalds 已提交
1184 1185 1186
}

static int
1187 1188 1189
same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
{
	return 0 == memcmp(v1->data, v2->data, sizeof(v1->data));
L
Linus Torvalds 已提交
1190 1191 1192
}

static int
1193 1194 1195
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
1196 1197
}

1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209
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;
}

1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
/*
 * 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);
}


1226
static bool
1227 1228
same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
{
1229
	if ((is_gss_cred(cr1) != is_gss_cred(cr2))
1230
		|| (cr1->cr_uid != cr2->cr_uid)
1231 1232 1233 1234 1235 1236 1237
		|| (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;
1238
	return 0 == strcmp(cr1->cr_principal, cr2->cr_principal);
L
Linus Torvalds 已提交
1239 1240
}

1241
static void gen_clid(struct nfs4_client *clp, struct nfsd_net *nn)
1242 1243 1244
{
	static u32 current_clientid = 1;

1245
	clp->cl_clientid.cl_boot = nn->boot_time;
L
Linus Torvalds 已提交
1246 1247 1248
	clp->cl_clientid.cl_id = current_clientid++; 
}

1249 1250
static void gen_confirm(struct nfs4_client *clp)
{
1251
	__be32 verf[2];
1252
	static u32 i;
L
Linus Torvalds 已提交
1253

1254 1255 1256
	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
1257 1258
}

1259
static struct nfs4_stid *find_stateid(struct nfs4_client *cl, stateid_t *t)
1260
{
1261
	return idr_find(&cl->cl_stateids, t->si_opaque.so_id);
J
J. Bruce Fields 已提交
1262 1263
}

1264
static struct nfs4_stid *find_stateid_by_type(struct nfs4_client *cl, stateid_t *t, char typemask)
1265 1266
{
	struct nfs4_stid *s;
J
J. Bruce Fields 已提交
1267

1268
	s = find_stateid(cl, t);
J
J. Bruce Fields 已提交
1269 1270
	if (!s)
		return NULL;
1271
	if (typemask & s->sc_type)
1272 1273 1274 1275
		return s;
	return NULL;
}

J
Jeff Layton 已提交
1276
static struct nfs4_client *create_client(struct xdr_netobj name,
1277 1278 1279 1280
		struct svc_rqst *rqstp, nfs4_verifier *verf)
{
	struct nfs4_client *clp;
	struct sockaddr *sa = svc_addr(rqstp);
1281
	int ret;
1282
	struct net *net = SVC_NET(rqstp);
1283
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
1284 1285 1286 1287 1288

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

1289
	INIT_LIST_HEAD(&clp->cl_sessions);
1290 1291
	ret = copy_cred(&clp->cl_cred, &rqstp->rq_cred);
	if (ret) {
1292
		spin_lock(&nn->client_lock);
1293
		free_client(clp);
1294
		spin_unlock(&nn->client_lock);
1295
		return NULL;
1296
	}
1297
	idr_init(&clp->cl_stateids);
1298
	atomic_set(&clp->cl_refcount, 0);
1299
	clp->cl_cb_state = NFSD4_CB_UNKNOWN;
1300 1301 1302 1303
	INIT_LIST_HEAD(&clp->cl_idhash);
	INIT_LIST_HEAD(&clp->cl_openowners);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_lru);
1304
	INIT_LIST_HEAD(&clp->cl_callbacks);
1305
	spin_lock_init(&clp->cl_lock);
1306
	nfsd4_init_callback(&clp->cl_cb_null);
B
Benny Halevy 已提交
1307
	clp->cl_time = get_seconds();
1308 1309 1310 1311 1312
	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);
1313
	clp->cl_cb_session = NULL;
1314
	clp->net = net;
1315 1316 1317
	return clp;
}

1318
static void
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 1356 1357 1358 1359
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
1360 1361
{
	unsigned int idhashval;
1362
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1363

1364
	clear_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
1365
	add_clp_to_name_tree(clp, &nn->unconf_name_tree);
L
Linus Torvalds 已提交
1366
	idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1367
	list_add(&clp->cl_idhash, &nn->unconf_id_hashtbl[idhashval]);
1368
	renew_client(clp);
L
Linus Torvalds 已提交
1369 1370
}

1371
static void
L
Linus Torvalds 已提交
1372 1373 1374
move_to_confirmed(struct nfs4_client *clp)
{
	unsigned int idhashval = clientid_hashval(clp->cl_clientid.cl_id);
1375
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
L
Linus Torvalds 已提交
1376 1377

	dprintk("NFSD: move_to_confirm nfs4_client %p\n", clp);
1378
	list_move(&clp->cl_idhash, &nn->conf_id_hashtbl[idhashval]);
1379
	rb_erase(&clp->cl_namenode, &nn->unconf_name_tree);
1380
	add_clp_to_name_tree(clp, &nn->conf_name_tree);
1381
	set_bit(NFSD4_CLIENT_CONFIRMED, &clp->cl_flags);
L
Linus Torvalds 已提交
1382 1383 1384 1385
	renew_client(clp);
}

static struct nfs4_client *
1386
find_confirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1387 1388 1389 1390
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1391
	list_for_each_entry(clp, &nn->conf_id_hashtbl[idhashval], cl_idhash) {
1392
		if (same_clid(&clp->cl_clientid, clid)) {
1393 1394
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
1395
			renew_client(clp);
L
Linus Torvalds 已提交
1396
			return clp;
1397
		}
L
Linus Torvalds 已提交
1398 1399 1400 1401 1402
	}
	return NULL;
}

static struct nfs4_client *
1403
find_unconfirmed_client(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
1404 1405 1406 1407
{
	struct nfs4_client *clp;
	unsigned int idhashval = clientid_hashval(clid->cl_id);

1408
	list_for_each_entry(clp, &nn->unconf_id_hashtbl[idhashval], cl_idhash) {
1409 1410 1411
		if (same_clid(&clp->cl_clientid, clid)) {
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
L
Linus Torvalds 已提交
1412
			return clp;
1413
		}
L
Linus Torvalds 已提交
1414 1415 1416 1417
	}
	return NULL;
}

1418
static bool clp_used_exchangeid(struct nfs4_client *clp)
1419
{
1420
	return clp->cl_exchange_flags != 0;
1421
} 
1422

1423
static struct nfs4_client *
1424
find_confirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1425
{
1426
	return find_clp_in_name_tree(name, &nn->conf_name_tree);
1427 1428 1429
}

static struct nfs4_client *
1430
find_unconfirmed_client_by_name(struct xdr_netobj *name, struct nfsd_net *nn)
1431
{
1432
	return find_clp_in_name_tree(name, &nn->unconf_name_tree);
1433 1434
}

1435
static void
1436
gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp)
L
Linus Torvalds 已提交
1437
{
1438
	struct nfs4_cb_conn *conn = &clp->cl_cb_conn;
1439 1440
	struct sockaddr	*sa = svc_addr(rqstp);
	u32 scopeid = rpc_get_scope_id(sa);
1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	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
1451 1452
		goto out_err;

1453
	conn->cb_addrlen = rpc_uaddr2sockaddr(clp->net, se->se_callback_addr_val,
1454
					    se->se_callback_addr_len,
1455 1456
					    (struct sockaddr *)&conn->cb_addr,
					    sizeof(conn->cb_addr));
1457

1458
	if (!conn->cb_addrlen || conn->cb_addr.ss_family != expected_family)
L
Linus Torvalds 已提交
1459
		goto out_err;
1460

1461 1462
	if (conn->cb_addr.ss_family == AF_INET6)
		((struct sockaddr_in6 *)&conn->cb_addr)->sin6_scope_id = scopeid;
1463

1464 1465
	conn->cb_prog = se->se_callback_prog;
	conn->cb_ident = se->se_callback_ident;
1466
	memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen);
L
Linus Torvalds 已提交
1467 1468
	return;
out_err:
1469 1470
	conn->cb_addr.ss_family = AF_UNSPEC;
	conn->cb_addrlen = 0;
N
Neil Brown 已提交
1471
	dprintk(KERN_INFO "NFSD: this client (clientid %08x/%08x) "
L
Linus Torvalds 已提交
1472 1473 1474 1475 1476 1477
		"will not receive delegations\n",
		clp->cl_clientid.cl_boot, clp->cl_clientid.cl_id);

	return;
}

1478
/*
1479
 * Cache a reply. nfsd4_check_drc_limit() has bounded the cache size.
1480 1481 1482 1483
 */
void
nfsd4_store_cache_entry(struct nfsd4_compoundres *resp)
{
1484 1485
	struct nfsd4_slot *slot = resp->cstate.slot;
	unsigned int base;
1486

1487
	dprintk("--> %s slot %p\n", __func__, slot);
1488

1489 1490
	slot->sl_opcnt = resp->opcnt;
	slot->sl_status = resp->cstate.status;
1491

1492
	slot->sl_flags |= NFSD4_SLOT_INITIALIZED;
1493
	if (nfsd4_not_cached(resp)) {
1494
		slot->sl_datalen = 0;
1495
		return;
1496
	}
1497 1498 1499 1500 1501 1502 1503
	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;
1504 1505 1506
}

/*
1507 1508 1509 1510
 * 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.
1511 1512
 *
 */
1513 1514 1515
static __be32
nfsd4_enc_sequence_replay(struct nfsd4_compoundargs *args,
			  struct nfsd4_compoundres *resp)
1516
{
1517 1518
	struct nfsd4_op *op;
	struct nfsd4_slot *slot = resp->cstate.slot;
1519

1520 1521 1522
	/* Encode the replayed sequence operation */
	op = &args->ops[resp->opcnt - 1];
	nfsd4_encode_operation(resp, op);
1523

1524
	/* Return nfserr_retry_uncached_rep in next operation. */
1525
	if (args->opcnt > 1 && !(slot->sl_flags & NFSD4_SLOT_CACHETHIS)) {
1526 1527 1528
		op = &args->ops[resp->opcnt++];
		op->status = nfserr_retry_uncached_rep;
		nfsd4_encode_operation(resp, op);
1529
	}
1530
	return op->status;
1531 1532 1533
}

/*
1534 1535
 * The sequence operation is not cached because we can use the slot and
 * session values.
1536 1537
 */
__be32
1538 1539
nfsd4_replay_cache_entry(struct nfsd4_compoundres *resp,
			 struct nfsd4_sequence *seq)
1540
{
1541
	struct nfsd4_slot *slot = resp->cstate.slot;
1542 1543
	__be32 status;

1544
	dprintk("--> %s slot %p\n", __func__, slot);
1545

1546 1547 1548 1549
	/* 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;
1550

1551 1552
	/* The sequence operation has been encoded, cstate->datap set. */
	memcpy(resp->cstate.datap, slot->sl_data, slot->sl_datalen);
1553

1554 1555 1556
	resp->opcnt = slot->sl_opcnt;
	resp->p = resp->cstate.datap + XDR_QUADLEN(slot->sl_datalen);
	status = slot->sl_status;
1557 1558 1559 1560

	return status;
}

A
Andy Adamson 已提交
1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
/*
 * 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;
}

1577 1578 1579 1580 1581 1582 1583 1584
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.
	 */
1585 1586 1587
	return !list_empty(&clp->cl_openowners)
		|| !list_empty(&clp->cl_delegations)
		|| !list_empty(&clp->cl_sessions);
1588 1589
}

A
Andy Adamson 已提交
1590 1591 1592 1593 1594
__be32
nfsd4_exchange_id(struct svc_rqst *rqstp,
		  struct nfsd4_compound_state *cstate,
		  struct nfsd4_exchange_id *exid)
{
A
Andy Adamson 已提交
1595
	struct nfs4_client *unconf, *conf, *new;
J
J. Bruce Fields 已提交
1596
	__be32 status;
1597
	char			addr_str[INET6_ADDRSTRLEN];
A
Andy Adamson 已提交
1598
	nfs4_verifier		verf = exid->verifier;
1599
	struct sockaddr		*sa = svc_addr(rqstp);
1600
	bool	update = exid->flags & EXCHGID4_FLAG_UPD_CONFIRMED_REC_A;
1601
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1602

1603
	rpc_ntop(sa, addr_str, sizeof(addr_str));
A
Andy Adamson 已提交
1604
	dprintk("%s rqstp=%p exid=%p clname.len=%u clname.data=%p "
1605
		"ip_addr=%s flags %x, spa_how %d\n",
A
Andy Adamson 已提交
1606
		__func__, rqstp, exid, exid->clname.len, exid->clname.data,
1607
		addr_str, exid->flags, exid->spa_how);
A
Andy Adamson 已提交
1608

1609
	if (exid->flags & ~EXCHGID4_FLAG_MASK_A)
A
Andy Adamson 已提交
1610 1611 1612 1613 1614 1615
		return nfserr_inval;

	/* Currently only support SP4_NONE */
	switch (exid->spa_how) {
	case SP4_NONE:
		break;
1616 1617
	default:				/* checked by xdr code */
		WARN_ON_ONCE(1);
A
Andy Adamson 已提交
1618 1619 1620 1621 1622
	case SP4_SSV:
	case SP4_MACH_CRED:
		return nfserr_serverfault;	/* no excuse :-/ */
	}

1623
	/* Cases below refer to rfc 5661 section 18.35.4: */
A
Andy Adamson 已提交
1624
	nfs4_lock_state();
1625
	conf = find_confirmed_client_by_name(&exid->clname, nn);
A
Andy Adamson 已提交
1626
	if (conf) {
1627 1628 1629
		bool creds_match = same_creds(&conf->cl_cred, &rqstp->rq_cred);
		bool verfs_match = same_verf(&verf, &conf->cl_verifier);

1630 1631
		if (update) {
			if (!clp_used_exchangeid(conf)) { /* buggy client */
1632
				status = nfserr_inval;
1633 1634
				goto out;
			}
1635
			if (!creds_match) { /* case 9 */
1636
				status = nfserr_perm;
1637 1638 1639
				goto out;
			}
			if (!verfs_match) { /* case 8 */
A
Andy Adamson 已提交
1640 1641 1642
				status = nfserr_not_same;
				goto out;
			}
1643 1644 1645 1646
			/* case 6 */
			exid->flags |= EXCHGID4_FLAG_CONFIRMED_R;
			new = conf;
			goto out_copy;
A
Andy Adamson 已提交
1647
		}
1648
		if (!creds_match) { /* case 3 */
1649 1650
			if (client_has_state(conf)) {
				status = nfserr_clid_inuse;
A
Andy Adamson 已提交
1651 1652 1653 1654 1655
				goto out;
			}
			expire_client(conf);
			goto out_new;
		}
1656
		if (verfs_match) { /* case 2 */
1657
			conf->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
1658 1659 1660 1661 1662
			new = conf;
			goto out_copy;
		}
		/* case 5, client reboot */
		goto out_new;
1663 1664
	}

1665
	if (update) { /* case 7 */
1666 1667
		status = nfserr_noent;
		goto out;
A
Andy Adamson 已提交
1668 1669
	}

1670
	unconf  = find_unconfirmed_client_by_name(&exid->clname, nn);
1671
	if (unconf) /* case 4, possible retry or client restart */
A
Andy Adamson 已提交
1672 1673
		expire_client(unconf);

1674
	/* case 1 (normal case) */
A
Andy Adamson 已提交
1675
out_new:
J
Jeff Layton 已提交
1676
	new = create_client(exid->clname, rqstp, &verf);
A
Andy Adamson 已提交
1677
	if (new == NULL) {
1678
		status = nfserr_jukebox;
A
Andy Adamson 已提交
1679 1680
		goto out;
	}
1681
	new->cl_minorversion = 1;
A
Andy Adamson 已提交
1682

1683
	gen_clid(new, nn);
1684
	add_to_unconfirmed(new);
A
Andy Adamson 已提交
1685 1686 1687 1688
out_copy:
	exid->clientid.cl_boot = new->cl_clientid.cl_boot;
	exid->clientid.cl_id = new->cl_clientid.cl_id;

1689
	exid->seqid = new->cl_cs_slot.sl_seqid + 1;
A
Andy Adamson 已提交
1690 1691 1692
	nfsd4_set_ex_flags(new, exid);

	dprintk("nfsd4_exchange_id seqid %d flags %x\n",
1693
		new->cl_cs_slot.sl_seqid, new->cl_exchange_flags);
A
Andy Adamson 已提交
1694 1695 1696 1697 1698
	status = nfs_ok;

out:
	nfs4_unlock_state();
	return status;
A
Andy Adamson 已提交
1699 1700
}

J
J. Bruce Fields 已提交
1701
static __be32
1702
check_slot_seqid(u32 seqid, u32 slot_seqid, int slot_inuse)
B
Benny Halevy 已提交
1703
{
1704 1705
	dprintk("%s enter. seqid %d slot_seqid %d\n", __func__, seqid,
		slot_seqid);
B
Benny Halevy 已提交
1706 1707

	/* The slot is in use, and no response has been sent. */
1708 1709
	if (slot_inuse) {
		if (seqid == slot_seqid)
B
Benny Halevy 已提交
1710 1711 1712 1713
			return nfserr_jukebox;
		else
			return nfserr_seq_misordered;
	}
1714
	/* Note unsigned 32-bit arithmetic handles wraparound: */
1715
	if (likely(seqid == slot_seqid + 1))
B
Benny Halevy 已提交
1716
		return nfs_ok;
1717
	if (seqid == slot_seqid)
B
Benny Halevy 已提交
1718 1719 1720 1721
		return nfserr_replay_cache;
	return nfserr_seq_misordered;
}

1722 1723 1724 1725 1726 1727 1728
/*
 * 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
1729
			   struct nfsd4_clid_slot *slot, __be32 nfserr)
1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
{
	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;
}

1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
#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
1760
static bool check_forechannel_attrs(struct nfsd4_channel_attrs fchannel)
1761 1762 1763 1764 1765
{
	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
1766 1767 1768 1769 1770
__be32
nfsd4_create_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_create_session *cr_ses)
{
1771
	struct sockaddr *sa = svc_addr(rqstp);
A
Andy Adamson 已提交
1772
	struct nfs4_client *conf, *unconf;
1773
	struct nfsd4_session *new;
1774
	struct nfsd4_conn *conn;
1775
	struct nfsd4_clid_slot *cs_slot = NULL;
J
J. Bruce Fields 已提交
1776
	__be32 status = 0;
1777
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
A
Andy Adamson 已提交
1778

1779 1780
	if (cr_ses->flags & ~SESSION4_FLAG_MASK_A)
		return nfserr_inval;
1781 1782
	if (check_forechannel_attrs(cr_ses->fore_channel))
		return nfserr_toosmall;
1783
	new = alloc_session(&cr_ses->fore_channel, nn);
1784 1785 1786 1787 1788 1789
	if (!new)
		return nfserr_jukebox;
	status = nfserr_jukebox;
	conn = alloc_conn_from_crses(rqstp, cr_ses);
	if (!conn)
		goto out_free_session;
1790

A
Andy Adamson 已提交
1791
	nfs4_lock_state();
1792
	unconf = find_unconfirmed_client(&cr_ses->clientid, true, nn);
1793
	conf = find_confirmed_client(&cr_ses->clientid, true, nn);
A
Andy Adamson 已提交
1794 1795

	if (conf) {
1796 1797
		cs_slot = &conf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1798
		if (status == nfserr_replay_cache) {
1799
			status = nfsd4_replay_create_session(cr_ses, cs_slot);
1800
			goto out_free_conn;
1801
		} else if (cr_ses->seqid != cs_slot->sl_seqid + 1) {
A
Andy Adamson 已提交
1802
			status = nfserr_seq_misordered;
1803
			goto out_free_conn;
A
Andy Adamson 已提交
1804 1805
		}
	} else if (unconf) {
J
J. Bruce Fields 已提交
1806
		struct nfs4_client *old;
A
Andy Adamson 已提交
1807
		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred) ||
1808
		    !rpc_cmp_addr(sa, (struct sockaddr *) &unconf->cl_addr)) {
A
Andy Adamson 已提交
1809
			status = nfserr_clid_inuse;
1810
			goto out_free_conn;
A
Andy Adamson 已提交
1811
		}
1812 1813
		cs_slot = &unconf->cl_cs_slot;
		status = check_slot_seqid(cr_ses->seqid, cs_slot->sl_seqid, 0);
1814 1815
		if (status) {
			/* an unconfirmed replay returns misordered */
A
Andy Adamson 已提交
1816
			status = nfserr_seq_misordered;
1817
			goto out_free_conn;
A
Andy Adamson 已提交
1818
		}
1819
		old = find_confirmed_client_by_name(&unconf->cl_name, nn);
J
J. Bruce Fields 已提交
1820 1821 1822
		if (old)
			expire_client(old);
		move_to_confirmed(unconf);
A
Andy Adamson 已提交
1823 1824 1825
		conf = unconf;
	} else {
		status = nfserr_stale_clientid;
1826
		goto out_free_conn;
A
Andy Adamson 已提交
1827
	}
1828
	status = nfs_ok;
1829 1830 1831 1832 1833 1834
	/*
	 * We do not support RDMA or persistent sessions
	 */
	cr_ses->flags &= ~SESSION4_PERSIST;
	cr_ses->flags &= ~SESSION4_RDMA;

1835 1836 1837
	init_session(rqstp, new, conf, cr_ses);
	nfsd4_init_conn(rqstp, conn, new);

1838
	memcpy(cr_ses->sessionid.data, new->se_sessionid.data,
A
Andy Adamson 已提交
1839
	       NFS4_MAX_SESSIONID_LEN);
1840 1841
	memcpy(&cr_ses->fore_channel, &new->se_fchannel,
		sizeof(struct nfsd4_channel_attrs));
1842
	cs_slot->sl_seqid++;
1843
	cr_ses->seqid = cs_slot->sl_seqid;
A
Andy Adamson 已提交
1844

1845 1846
	/* 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
1847 1848 1849 1850
out:
	nfs4_unlock_state();
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
1851 1852 1853 1854 1855
out_free_conn:
	free_conn(conn);
out_free_session:
	__free_session(new);
	goto out;
A
Andy Adamson 已提交
1856 1857
}

1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
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;
}

1872 1873 1874
__be32 nfsd4_backchannel_ctl(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, struct nfsd4_backchannel_ctl *bc)
{
	struct nfsd4_session *session = cstate->session;
1875
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1876

1877
	spin_lock(&nn->client_lock);
1878 1879
	session->se_cb_prog = bc->bc_cb_program;
	session->se_cb_sec = bc->bc_cb_sec;
1880
	spin_unlock(&nn->client_lock);
1881 1882 1883 1884 1885 1886

	nfsd4_probe_callback(session->se_client);

	return nfs_ok;
}

1887 1888 1889 1890 1891
__be32 nfsd4_bind_conn_to_session(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
		     struct nfsd4_bind_conn_to_session *bcts)
{
	__be32 status;
1892
	struct nfsd4_conn *conn;
1893
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
1894 1895 1896

	if (!nfsd4_last_compound_op(rqstp))
		return nfserr_not_only_op;
1897
	spin_lock(&nn->client_lock);
1898
	cstate->session = find_in_sessionid_hashtbl(&bcts->sessionid, SVC_NET(rqstp));
1899 1900 1901 1902 1903 1904
	/* 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);
	}
1905
	spin_unlock(&nn->client_lock);
1906 1907 1908 1909
	if (!cstate->session)
		return nfserr_badsession;

	status = nfsd4_map_bcts_dir(&bcts->dir);
1910 1911 1912 1913 1914
	if (status)
		return status;
	conn = alloc_conn(rqstp, bcts->dir);
	if (!conn)
		return nfserr_jukebox;
1915
	nfsd4_init_conn(rqstp, conn, cstate->session);
1916
	return nfs_ok;
1917 1918
}

1919 1920 1921 1922 1923 1924 1925
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
1926 1927 1928 1929 1930
__be32
nfsd4_destroy_session(struct svc_rqst *r,
		      struct nfsd4_compound_state *cstate,
		      struct nfsd4_destroy_session *sessionid)
{
B
Benny Halevy 已提交
1931
	struct nfsd4_session *ses;
J
J. Bruce Fields 已提交
1932
	__be32 status = nfserr_badsession;
1933
	struct nfsd_net *nn = net_generic(SVC_NET(r), nfsd_net_id);
B
Benny Halevy 已提交
1934 1935 1936 1937 1938 1939 1940 1941

	/* 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?
	 */

1942
	if (nfsd4_compound_in_session(cstate->session, &sessionid->sessionid)) {
1943 1944 1945
		if (!nfsd4_last_compound_op(r))
			return nfserr_not_only_op;
	}
B
Benny Halevy 已提交
1946
	dump_sessionid(__func__, &sessionid->sessionid);
1947
	spin_lock(&nn->client_lock);
1948
	ses = find_in_sessionid_hashtbl(&sessionid->sessionid, SVC_NET(r));
B
Benny Halevy 已提交
1949
	if (!ses) {
1950
		spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1951 1952 1953 1954
		goto out;
	}

	unhash_session(ses);
1955
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1956

1957
	nfs4_lock_state();
1958
	nfsd4_probe_callback_sync(ses->se_client);
1959
	nfs4_unlock_state();
1960

1961
	spin_lock(&nn->client_lock);
1962
	nfsd4_del_conns(ses);
1963
	nfsd4_put_session_locked(ses);
1964
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
1965 1966 1967 1968
	status = nfs_ok;
out:
	dprintk("%s returns %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
1969 1970
}

1971
static struct nfsd4_conn *__nfsd4_find_conn(struct svc_xprt *xpt, struct nfsd4_session *s)
1972 1973 1974 1975
{
	struct nfsd4_conn *c;

	list_for_each_entry(c, &s->se_conns, cn_persession) {
1976
		if (c->cn_xprt == xpt) {
1977 1978 1979 1980 1981 1982
			return c;
		}
	}
	return NULL;
}

1983
static void nfsd4_sequence_check_conn(struct nfsd4_conn *new, struct nfsd4_session *ses)
1984 1985
{
	struct nfs4_client *clp = ses->se_client;
1986
	struct nfsd4_conn *c;
1987
	int ret;
1988 1989

	spin_lock(&clp->cl_lock);
1990
	c = __nfsd4_find_conn(new->cn_xprt, ses);
1991 1992 1993 1994 1995 1996 1997
	if (c) {
		spin_unlock(&clp->cl_lock);
		free_conn(new);
		return;
	}
	__nfsd4_hash_conn(new, ses);
	spin_unlock(&clp->cl_lock);
1998 1999 2000 2001
	ret = nfsd4_register_conn(new);
	if (ret)
		/* oops; xprt is already down: */
		nfsd4_conn_lost(&new->cn_xpt_user);
2002 2003 2004
	return;
}

2005 2006 2007 2008 2009 2010 2011
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
2012 2013 2014 2015 2016 2017 2018 2019
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
2020
__be32
B
Benny Halevy 已提交
2021
nfsd4_sequence(struct svc_rqst *rqstp,
A
Andy Adamson 已提交
2022 2023 2024
	       struct nfsd4_compound_state *cstate,
	       struct nfsd4_sequence *seq)
{
2025
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
B
Benny Halevy 已提交
2026 2027
	struct nfsd4_session *session;
	struct nfsd4_slot *slot;
2028
	struct nfsd4_conn *conn;
J
J. Bruce Fields 已提交
2029
	__be32 status;
2030
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
B
Benny Halevy 已提交
2031

2032 2033 2034
	if (resp->opcnt != 1)
		return nfserr_sequence_pos;

2035 2036 2037 2038 2039 2040 2041 2042
	/*
	 * Will be either used or freed by nfsd4_sequence_check_conn
	 * below.
	 */
	conn = alloc_conn(rqstp, NFS4_CDFC4_FORE);
	if (!conn)
		return nfserr_jukebox;

2043
	spin_lock(&nn->client_lock);
B
Benny Halevy 已提交
2044
	status = nfserr_badsession;
2045
	session = find_in_sessionid_hashtbl(&seq->sessionid, SVC_NET(rqstp));
B
Benny Halevy 已提交
2046 2047 2048
	if (!session)
		goto out;

2049 2050 2051 2052
	status = nfserr_too_many_ops;
	if (nfsd4_session_too_many_ops(rqstp, session))
		goto out;

M
Mi Jinlong 已提交
2053 2054 2055 2056
	status = nfserr_req_too_big;
	if (nfsd4_request_too_big(rqstp, session))
		goto out;

B
Benny Halevy 已提交
2057
	status = nfserr_badslot;
2058
	if (seq->slotid >= session->se_fchannel.maxreqs)
B
Benny Halevy 已提交
2059 2060
		goto out;

2061
	slot = session->se_slots[seq->slotid];
B
Benny Halevy 已提交
2062 2063
	dprintk("%s: slotid %d\n", __func__, seq->slotid);

2064 2065 2066 2067 2068
	/* 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;

2069 2070
	status = check_slot_seqid(seq->seqid, slot->sl_seqid,
					slot->sl_flags & NFSD4_SLOT_INUSE);
B
Benny Halevy 已提交
2071
	if (status == nfserr_replay_cache) {
2072 2073 2074
		status = nfserr_seq_misordered;
		if (!(slot->sl_flags & NFSD4_SLOT_INITIALIZED))
			goto out;
B
Benny Halevy 已提交
2075 2076
		cstate->slot = slot;
		cstate->session = session;
A
Andy Adamson 已提交
2077
		/* Return the cached reply status and set cstate->status
2078
		 * for nfsd4_proc_compound processing */
2079
		status = nfsd4_replay_cache_entry(resp, seq);
A
Andy Adamson 已提交
2080
		cstate->status = nfserr_replay_cache;
2081
		goto out;
B
Benny Halevy 已提交
2082 2083 2084 2085
	}
	if (status)
		goto out;

2086 2087
	nfsd4_sequence_check_conn(conn, session);
	conn = NULL;
2088

B
Benny Halevy 已提交
2089 2090
	/* Success! bump slot seqid */
	slot->sl_seqid = seq->seqid;
2091
	slot->sl_flags |= NFSD4_SLOT_INUSE;
2092 2093
	if (seq->cachethis)
		slot->sl_flags |= NFSD4_SLOT_CACHETHIS;
2094 2095
	else
		slot->sl_flags &= ~NFSD4_SLOT_CACHETHIS;
B
Benny Halevy 已提交
2096 2097 2098 2099 2100

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

out:
2101
	/* Hold a session reference until done processing the compound. */
2102
	if (cstate->session) {
2103 2104
		struct nfs4_client *clp = session->se_client;

2105
		nfsd4_get_session(cstate->session);
2106
		atomic_inc(&clp->cl_refcount);
2107 2108
		switch (clp->cl_cb_state) {
		case NFSD4_CB_DOWN:
2109
			seq->status_flags = SEQ4_STATUS_CB_PATH_DOWN;
2110 2111
			break;
		case NFSD4_CB_FAULT:
2112
			seq->status_flags = SEQ4_STATUS_BACKCHANNEL_FAULT;
2113
			break;
2114 2115
		default:
			seq->status_flags = 0;
2116
		}
2117
	}
2118
	kfree(conn);
2119
	spin_unlock(&nn->client_lock);
B
Benny Halevy 已提交
2120 2121
	dprintk("%s: return %d\n", __func__, ntohl(status));
	return status;
A
Andy Adamson 已提交
2122 2123
}

2124 2125 2126 2127
__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
2128
	__be32 status = 0;
2129
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
2130 2131

	nfs4_lock_state();
2132
	unconf = find_unconfirmed_client(&dc->clientid, true, nn);
2133
	conf = find_confirmed_client(&dc->clientid, true, nn);
2134 2135 2136 2137

	if (conf) {
		clp = conf;

2138
		if (!is_client_expired(conf) && client_has_state(conf)) {
2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
			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;
}

2162 2163 2164
__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
2165
	__be32 status = 0;
2166

2167 2168 2169 2170 2171 2172 2173 2174 2175
	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;
	}
2176

2177
	nfs4_lock_state();
2178
	status = nfserr_complete_already;
2179 2180
	if (test_and_set_bit(NFSD4_CLIENT_RECLAIM_COMPLETE,
			     &cstate->session->se_client->cl_flags))
2181 2182 2183 2184
		goto out;

	status = nfserr_stale_clientid;
	if (is_client_expired(cstate->session->se_client))
2185 2186 2187 2188 2189 2190 2191
		/*
		 * 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.
		 */
2192 2193 2194
		goto out;

	status = nfs_ok;
2195
	nfsd4_client_record_create(cstate->session->se_client);
2196
out:
2197
	nfs4_unlock_state();
2198
	return status;
2199 2200
}

2201
__be32
2202 2203
nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_setclientid *setclid)
L
Linus Torvalds 已提交
2204
{
2205
	struct xdr_netobj 	clname = setclid->se_name;
L
Linus Torvalds 已提交
2206
	nfs4_verifier		clverifier = setclid->se_verf;
2207
	struct nfs4_client	*conf, *unconf, *new;
2208
	__be32 			status;
2209 2210
	struct nfsd_net		*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

2211
	/* Cases below refer to rfc 3530 section 14.2.33: */
L
Linus Torvalds 已提交
2212
	nfs4_lock_state();
2213
	conf = find_confirmed_client_by_name(&clname, nn);
2214
	if (conf) {
2215
		/* case 0: */
L
Linus Torvalds 已提交
2216
		status = nfserr_clid_inuse;
2217 2218
		if (clp_used_exchangeid(conf))
			goto out;
2219
		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)) {
2220 2221 2222 2223 2224
			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
2225 2226 2227
			goto out;
		}
	}
2228
	unconf = find_unconfirmed_client_by_name(&clname, nn);
2229 2230
	if (unconf)
		expire_client(unconf);
2231
	status = nfserr_jukebox;
J
Jeff Layton 已提交
2232
	new = create_client(clname, rqstp, &clverifier);
2233 2234
	if (new == NULL)
		goto out;
2235
	if (conf && same_verf(&conf->cl_verifier, &clverifier))
2236
		/* case 1: probable callback update */
L
Linus Torvalds 已提交
2237
		copy_clid(new, conf);
2238
	else /* case 4 (new client) or cases 2, 3 (client reboot): */
2239
		gen_clid(new, nn);
2240
	new->cl_minorversion = 0;
2241
	gen_callback(new, setclid, rqstp);
2242
	add_to_unconfirmed(new);
L
Linus Torvalds 已提交
2243 2244 2245 2246 2247 2248 2249 2250 2251 2252
	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;
}


2253
__be32
2254 2255 2256
nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
			 struct nfsd4_compound_state *cstate,
			 struct nfsd4_setclientid_confirm *setclientid_confirm)
L
Linus Torvalds 已提交
2257
{
2258
	struct nfs4_client *conf, *unconf;
L
Linus Torvalds 已提交
2259 2260
	nfs4_verifier confirm = setclientid_confirm->sc_confirm; 
	clientid_t * clid = &setclientid_confirm->sc_clientid;
2261
	__be32 status;
2262
	struct nfsd_net	*nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
2263

2264
	if (STALE_CLIENTID(clid, nn))
L
Linus Torvalds 已提交
2265 2266
		return nfserr_stale_clientid;
	nfs4_lock_state();
2267

2268
	conf = find_confirmed_client(clid, false, nn);
2269
	unconf = find_unconfirmed_client(clid, false, nn);
2270
	/*
2271 2272 2273 2274
	 * 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.
2275
	 */
2276 2277 2278 2279 2280
	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;
2281
	/* cases below refer to rfc 3530 section 14.2.34: */
2282 2283
	if (!unconf || !same_verf(&confirm, &unconf->cl_confirm)) {
		if (conf && !unconf) /* case 2: probable retransmit */
L
Linus Torvalds 已提交
2284
			status = nfs_ok;
2285 2286 2287 2288 2289 2290
		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 */
2291 2292 2293
		nfsd4_change_callback(conf, &unconf->cl_cb_conn);
		nfsd4_probe_callback(conf);
		expire_client(unconf);
2294
	} else { /* case 3: normal case; new or rebooted client */
2295
		conf = find_confirmed_client_by_name(&unconf->cl_name, nn);
2296
		if (conf)
2297 2298
			expire_client(conf);
		move_to_confirmed(unconf);
2299
		nfsd4_probe_callback(unconf);
2300
	}
L
Linus Torvalds 已提交
2301 2302 2303 2304 2305
out:
	nfs4_unlock_state();
	return status;
}

2306 2307 2308 2309 2310
static struct nfs4_file *nfsd4_alloc_file(void)
{
	return kmem_cache_alloc(file_slab, GFP_KERNEL);
}

L
Linus Torvalds 已提交
2311
/* OPEN Share state helper functions */
2312
static void nfsd4_init_file(struct nfs4_file *fp, struct inode *ino)
L
Linus Torvalds 已提交
2313 2314 2315
{
	unsigned int hashval = file_hashval(ino);

2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
	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
2328 2329
}

N
NeilBrown 已提交
2330
static void
2331
nfsd4_free_slab(struct kmem_cache **slab)
L
Linus Torvalds 已提交
2332
{
N
NeilBrown 已提交
2333 2334
	if (*slab == NULL)
		return;
2335
	kmem_cache_destroy(*slab);
N
NeilBrown 已提交
2336
	*slab = NULL;
L
Linus Torvalds 已提交
2337 2338
}

2339
void
L
Linus Torvalds 已提交
2340 2341
nfsd4_free_slabs(void)
{
2342 2343
	nfsd4_free_slab(&openowner_slab);
	nfsd4_free_slab(&lockowner_slab);
N
NeilBrown 已提交
2344
	nfsd4_free_slab(&file_slab);
N
NeilBrown 已提交
2345
	nfsd4_free_slab(&stateid_slab);
N
NeilBrown 已提交
2346
	nfsd4_free_slab(&deleg_slab);
N
NeilBrown 已提交
2347
}
L
Linus Torvalds 已提交
2348

2349
int
N
NeilBrown 已提交
2350 2351
nfsd4_init_slabs(void)
{
2352 2353 2354 2355 2356
	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",
2357
			sizeof(struct nfs4_lockowner), 0, 0, NULL);
2358
	if (lockowner_slab == NULL)
N
NeilBrown 已提交
2359 2360
		goto out_nomem;
	file_slab = kmem_cache_create("nfsd4_files",
2361
			sizeof(struct nfs4_file), 0, 0, NULL);
N
NeilBrown 已提交
2362 2363
	if (file_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2364
	stateid_slab = kmem_cache_create("nfsd4_stateids",
2365
			sizeof(struct nfs4_ol_stateid), 0, 0, NULL);
N
NeilBrown 已提交
2366 2367
	if (stateid_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2368
	deleg_slab = kmem_cache_create("nfsd4_delegations",
2369
			sizeof(struct nfs4_delegation), 0, 0, NULL);
N
NeilBrown 已提交
2370 2371
	if (deleg_slab == NULL)
		goto out_nomem;
N
NeilBrown 已提交
2372 2373 2374 2375 2376
	return 0;
out_nomem:
	nfsd4_free_slabs();
	dprintk("nfsd4: out of memory while initializing nfsv4\n");
	return -ENOMEM;
L
Linus Torvalds 已提交
2377 2378
}

2379 2380 2381 2382 2383 2384 2385
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
2386
{
2387 2388
	kfree(lo->lo_owner.so_owner.data);
	kmem_cache_free(lockowner_slab, lo);
L
Linus Torvalds 已提交
2389 2390
}

2391
static void init_nfs4_replay(struct nfs4_replay *rp)
L
Linus Torvalds 已提交
2392
{
2393 2394 2395
	rp->rp_status = nfserr_serverfault;
	rp->rp_buflen = 0;
	rp->rp_buf = rp->rp_ibuf;
L
Linus Torvalds 已提交
2396 2397
}

2398
static inline void *alloc_stateowner(struct kmem_cache *slab, struct xdr_netobj *owner, struct nfs4_client *clp)
2399
{
L
Linus Torvalds 已提交
2400 2401
	struct nfs4_stateowner *sop;

2402
	sop = kmem_cache_alloc(slab, GFP_KERNEL);
2403 2404 2405 2406 2407
	if (!sop)
		return NULL;

	sop->so_owner.data = kmemdup(owner->data, owner->len, GFP_KERNEL);
	if (!sop->so_owner.data) {
2408
		kmem_cache_free(slab, sop);
L
Linus Torvalds 已提交
2409
		return NULL;
2410 2411 2412
	}
	sop->so_owner.len = owner->len;

2413
	INIT_LIST_HEAD(&sop->so_stateids);
2414 2415 2416 2417 2418
	sop->so_client = clp;
	init_nfs4_replay(&sop->so_replay);
	return sop;
}

2419
static void hash_openowner(struct nfs4_openowner *oo, struct nfs4_client *clp, unsigned int strhashval)
2420
{
2421 2422 2423
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);

	list_add(&oo->oo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
2424
	list_add(&oo->oo_perclient, &clp->cl_openowners);
2425 2426
}

2427
static struct nfs4_openowner *
2428
alloc_init_open_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfsd4_open *open) {
2429
	struct nfs4_openowner *oo;
2430

2431 2432
	oo = alloc_stateowner(openowner_slab, &open->op_owner, clp);
	if (!oo)
2433
		return NULL;
2434 2435
	oo->oo_owner.so_is_open_owner = 1;
	oo->oo_owner.so_seqid = open->op_seqid;
2436
	oo->oo_flags = NFS4_OO_NEW;
2437
	oo->oo_time = 0;
2438
	oo->oo_last_closed_stid = NULL;
2439 2440 2441
	INIT_LIST_HEAD(&oo->oo_close_lru);
	hash_openowner(oo, clp, strhashval);
	return oo;
L
Linus Torvalds 已提交
2442 2443
}

2444
static void init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
2445
	struct nfs4_openowner *oo = open->op_openowner;
2446
	struct nfs4_client *clp = oo->oo_owner.so_client;
L
Linus Torvalds 已提交
2447

2448
	init_stid(&stp->st_stid, clp, NFS4_OPEN_STID);
2449
	INIT_LIST_HEAD(&stp->st_lockowners);
2450
	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
2451
	list_add(&stp->st_perfile, &fp->fi_stateids);
2452
	stp->st_stateowner = &oo->oo_owner;
2453
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
2454 2455 2456
	stp->st_file = fp;
	stp->st_access_bmap = 0;
	stp->st_deny_bmap = 0;
2457
	set_access(open->op_share_access, stp);
2458
	set_deny(open->op_share_deny, stp);
2459
	stp->st_openstp = NULL;
L
Linus Torvalds 已提交
2460 2461
}

2462
static void
2463
move_to_close_lru(struct nfs4_openowner *oo, struct net *net)
L
Linus Torvalds 已提交
2464
{
2465 2466
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);

2467
	dprintk("NFSD: move_to_close_lru nfs4_openowner %p\n", oo);
L
Linus Torvalds 已提交
2468

2469
	list_move_tail(&oo->oo_close_lru, &nn->close_lru);
2470
	oo->oo_time = get_seconds();
L
Linus Torvalds 已提交
2471 2472 2473
}

static int
2474 2475 2476 2477 2478 2479
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
2480 2481
}

2482
static struct nfs4_openowner *
2483 2484
find_openstateowner_str(unsigned int hashval, struct nfsd4_open *open,
			bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2485
{
2486 2487
	struct nfs4_stateowner *so;
	struct nfs4_openowner *oo;
2488
	struct nfs4_client *clp;
L
Linus Torvalds 已提交
2489

2490
	list_for_each_entry(so, &nn->ownerstr_hashtbl[hashval], so_strhash) {
2491 2492
		if (!so->so_is_open_owner)
			continue;
2493 2494
		if (same_owner_str(so, &open->op_owner, &open->op_clientid)) {
			oo = openowner(so);
2495 2496 2497
			clp = oo->oo_owner.so_client;
			if ((bool)clp->cl_minorversion != sessions)
				return NULL;
2498 2499 2500
			renew_client(oo->oo_owner.so_client);
			return oo;
		}
L
Linus Torvalds 已提交
2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511
	}
	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;

2512
	spin_lock(&recall_lock);
L
Linus Torvalds 已提交
2513
	list_for_each_entry(fp, &file_hashtbl[hashval], fi_hash) {
2514 2515
		if (fp->fi_inode == ino) {
			get_nfs4_file(fp);
2516
			spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2517
			return fp;
2518
		}
L
Linus Torvalds 已提交
2519
	}
2520
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2521 2522 2523 2524 2525 2526 2527
	return NULL;
}

/*
 * Called to check deny when READ with all zero stateid or
 * WRITE with all zero or all one stateid
 */
2528
static __be32
L
Linus Torvalds 已提交
2529 2530 2531 2532
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;
2533
	struct nfs4_ol_stateid *stp;
2534
	__be32 ret;
L
Linus Torvalds 已提交
2535 2536 2537 2538

	dprintk("NFSD: nfs4_share_conflict\n");

	fp = find_file(ino);
2539 2540
	if (!fp)
		return nfs_ok;
2541
	ret = nfserr_locked;
L
Linus Torvalds 已提交
2542
	/* Search for conflicting share reservations */
2543
	list_for_each_entry(stp, &fp->fi_stateids, st_perfile) {
2544 2545
		if (test_deny(deny_type, stp) ||
		    test_deny(NFS4_SHARE_DENY_BOTH, stp))
2546
			goto out;
L
Linus Torvalds 已提交
2547
	}
2548 2549 2550 2551
	ret = nfs_ok;
out:
	put_nfs4_file(fp);
	return ret;
L
Linus Torvalds 已提交
2552 2553
}

2554
static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
L
Linus Torvalds 已提交
2555 2556 2557 2558 2559 2560 2561 2562 2563 2564
{
	/* 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);

2565
	/* only place dl_time is set. protected by lock_flocks*/
L
Linus Torvalds 已提交
2566 2567
	dp->dl_time = get_seconds();

2568 2569 2570
	nfsd4_cb_recall(dp);
}

2571
/* Called from break_lease() with lock_flocks() held. */
2572 2573
static void nfsd_break_deleg_cb(struct file_lock *fl)
{
2574 2575
	struct nfs4_file *fp = (struct nfs4_file *)fl->fl_owner;
	struct nfs4_delegation *dp;
2576

2577 2578 2579 2580 2581 2582 2583 2584
	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;
	}
2585 2586
	/*
	 * We don't want the locks code to timeout the lease for us;
2587
	 * we'll remove it ourself if a delegation isn't returned
2588
	 * in time:
2589 2590
	 */
	fl->fl_break_time = 0;
L
Linus Torvalds 已提交
2591

2592
	spin_lock(&recall_lock);
2593 2594 2595
	fp->fi_had_conflict = true;
	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
		nfsd_break_one_deleg(dp);
2596
	spin_unlock(&recall_lock);
L
Linus Torvalds 已提交
2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607
}

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

2608
static const struct lock_manager_operations nfsd_lease_mng_ops = {
J
J. Bruce Fields 已提交
2609 2610
	.lm_break = nfsd_break_deleg_cb,
	.lm_change = nfsd_change_deleg_cb,
L
Linus Torvalds 已提交
2611 2612
};

2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
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
2623

2624
__be32
A
Andy Adamson 已提交
2625
nfsd4_process_open1(struct nfsd4_compound_state *cstate,
2626
		    struct nfsd4_open *open, struct nfsd_net *nn)
L
Linus Torvalds 已提交
2627 2628 2629 2630
{
	clientid_t *clientid = &open->op_clientid;
	struct nfs4_client *clp = NULL;
	unsigned int strhashval;
2631
	struct nfs4_openowner *oo = NULL;
2632
	__be32 status;
L
Linus Torvalds 已提交
2633

2634
	if (STALE_CLIENTID(&open->op_clientid, nn))
L
Linus Torvalds 已提交
2635
		return nfserr_stale_clientid;
2636 2637 2638 2639 2640 2641 2642
	/*
	 * 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
2643

2644
	strhashval = ownerstr_hashval(clientid->cl_id, &open->op_owner);
2645
	oo = find_openstateowner_str(strhashval, open, cstate->minorversion, nn);
2646 2647
	open->op_openowner = oo;
	if (!oo) {
2648 2649
		clp = find_confirmed_client(clientid, cstate->minorversion,
					    nn);
L
Linus Torvalds 已提交
2650
		if (clp == NULL)
2651
			return nfserr_expired;
2652
		goto new_owner;
L
Linus Torvalds 已提交
2653
	}
2654
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED)) {
2655
		/* Replace unconfirmed owners without checking for replay. */
2656 2657 2658
		clp = oo->oo_owner.so_client;
		release_openowner(oo);
		open->op_openowner = NULL;
2659
		goto new_owner;
2660
	}
2661 2662 2663 2664 2665
	status = nfsd4_check_seqid(cstate, &oo->oo_owner, open->op_seqid);
	if (status)
		return status;
	clp = oo->oo_owner.so_client;
	goto alloc_stateid;
2666 2667 2668 2669 2670
new_owner:
	oo = alloc_init_open_stateowner(strhashval, clp, open);
	if (oo == NULL)
		return nfserr_jukebox;
	open->op_openowner = oo;
2671 2672 2673 2674
alloc_stateid:
	open->op_stp = nfs4_alloc_stateid(clp);
	if (!open->op_stp)
		return nfserr_jukebox;
2675
	return nfs_ok;
L
Linus Torvalds 已提交
2676 2677
}

2678
static inline __be32
N
NeilBrown 已提交
2679 2680 2681 2682 2683 2684 2685 2686
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;
}

2687
static int share_access_to_flags(u32 share_access)
2688
{
2689
	return share_access == NFS4_SHARE_ACCESS_READ ? RD_STATE : WR_STATE;
2690 2691
}

2692
static struct nfs4_delegation *find_deleg_stateid(struct nfs4_client *cl, stateid_t *s)
2693
{
2694
	struct nfs4_stid *ret;
2695

2696
	ret = find_stateid_by_type(cl, s, NFS4_DELEG_STID);
2697 2698 2699
	if (!ret)
		return NULL;
	return delegstateid(ret);
2700 2701
}

2702 2703 2704 2705 2706 2707
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;
}

2708
static __be32
2709
nfs4_check_deleg(struct nfs4_client *cl, struct nfs4_file *fp, struct nfsd4_open *open,
2710 2711 2712
		struct nfs4_delegation **dp)
{
	int flags;
2713
	__be32 status = nfserr_bad_stateid;
2714

2715
	*dp = find_deleg_stateid(cl, &open->op_delegate_stateid);
2716
	if (*dp == NULL)
2717
		goto out;
2718
	flags = share_access_to_flags(open->op_share_access);
2719 2720 2721
	status = nfs4_check_delegmode(*dp, flags);
	if (status)
		*dp = NULL;
2722
out:
2723
	if (!nfsd4_is_deleg_cur(open))
2724 2725 2726
		return nfs_ok;
	if (status)
		return status;
2727
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
2728
	return nfs_ok;
2729 2730
}

2731
static __be32
2732
nfs4_check_open(struct nfs4_file *fp, struct nfsd4_open *open, struct nfs4_ol_stateid **stpp)
L
Linus Torvalds 已提交
2733
{
2734
	struct nfs4_ol_stateid *local;
2735
	struct nfs4_openowner *oo = open->op_openowner;
L
Linus Torvalds 已提交
2736

2737
	list_for_each_entry(local, &fp->fi_stateids, st_perfile) {
L
Linus Torvalds 已提交
2738 2739 2740 2741
		/* ignore lock owners */
		if (local->st_stateowner->so_is_open_owner == 0)
			continue;
		/* remember if we have seen this open owner */
2742
		if (local->st_stateowner == &oo->oo_owner)
L
Linus Torvalds 已提交
2743 2744 2745
			*stpp = local;
		/* check for conflicting share reservations */
		if (!test_share(local, open))
2746
			return nfserr_share_denied;
L
Linus Torvalds 已提交
2747
	}
2748
	return nfs_ok;
L
Linus Torvalds 已提交
2749 2750
}

2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761
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;
}

2762 2763
static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
		struct svc_fh *cur_fh, struct nfsd4_open *open)
2764 2765
{
	__be32 status;
2766 2767 2768
	int oflag = nfs4_access_to_omode(open->op_share_access);
	int access = nfs4_access_to_access(open->op_share_access);

2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779
	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;
}

2780
static inline __be32
L
Linus Torvalds 已提交
2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
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))
2791
		return nfserr_inval;
L
Linus Torvalds 已提交
2792 2793 2794
	return nfsd_setattr(rqstp, fh, &iattr, 0, (time_t)0);
}

2795
static __be32
2796
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
2797
{
2798
	u32 op_share_access = open->op_share_access;
2799
	bool new_access;
2800
	__be32 status;
L
Linus Torvalds 已提交
2801

2802
	new_access = !test_access(op_share_access, stp);
2803
	if (new_access) {
2804
		status = nfs4_get_vfs_file(rqstp, fp, cur_fh, open);
2805 2806
		if (status)
			return status;
2807
	}
L
Linus Torvalds 已提交
2808 2809
	status = nfsd4_truncate(rqstp, cur_fh, open);
	if (status) {
2810
		if (new_access) {
2811
			int oflag = nfs4_access_to_omode(op_share_access);
2812 2813
			nfs4_file_put_access(fp, oflag);
		}
L
Linus Torvalds 已提交
2814 2815 2816
		return status;
	}
	/* remember the open */
2817
	set_access(op_share_access, stp);
2818
	set_deny(open->op_share_deny, stp);
L
Linus Torvalds 已提交
2819 2820 2821 2822 2823 2824

	return nfs_ok;
}


static void
2825
nfs4_set_claim_prev(struct nfsd4_open *open, bool has_session)
L
Linus Torvalds 已提交
2826
{
2827
	open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
L
Linus Torvalds 已提交
2828 2829
}

2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842
/* 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;
}

2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854
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;
2855
	fl->fl_owner = (fl_owner_t)(dp->dl_file);
2856 2857 2858 2859
	fl->fl_pid = current->tgid;
	return fl;
}

2860 2861
static int nfs4_setlease(struct nfs4_delegation *dp, int flag)
{
2862
	struct nfs4_file *fp = dp->dl_file;
2863 2864 2865 2866 2867 2868
	struct file_lock *fl;
	int status;

	fl = nfs4_alloc_init_lease(dp, flag);
	if (!fl)
		return -ENOMEM;
2869
	fl->fl_file = find_readable_file(fp);
2870
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2871
	status = vfs_setlease(fl->fl_file, fl->fl_type, &fl);
2872
	if (status) {
2873
		list_del_init(&dp->dl_perclnt);
2874 2875 2876
		locks_free_lock(fl);
		return -ENOMEM;
	}
2877
	fp->fi_lease = fl;
A
Al Viro 已提交
2878
	fp->fi_deleg_file = get_file(fl->fl_file);
2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897
	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);
2898
	list_add(&dp->dl_perclnt, &dp->dl_stid.sc_client->cl_delegations);
2899 2900 2901
	return 0;
}

2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917
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:
2918
			WARN_ON_ONCE(1);
2919 2920 2921 2922
		}
	}
}

L
Linus Torvalds 已提交
2923 2924 2925 2926
/*
 * Attempt to hand out a delegation.
 */
static void
2927 2928
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
2929 2930
{
	struct nfs4_delegation *dp;
2931
	struct nfs4_openowner *oo = container_of(stp->st_stateowner, struct nfs4_openowner, oo_owner);
2932
	int cb_up;
2933
	int status = 0, flag = 0;
L
Linus Torvalds 已提交
2934

2935
	cb_up = nfsd4_cb_channel_good(oo->oo_owner.so_client);
L
Linus Torvalds 已提交
2936
	flag = NFS4_OPEN_DELEGATE_NONE;
2937 2938 2939
	open->op_recall = 0;
	switch (open->op_claim_type) {
		case NFS4_OPEN_CLAIM_PREVIOUS:
2940
			if (!cb_up)
2941 2942 2943 2944 2945 2946 2947 2948
				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.... */
2949
			if (locks_in_grace(net))
2950
				goto out;
2951
			if (!cb_up || !(oo->oo_flags & NFS4_OO_CONFIRMED))
2952 2953 2954 2955 2956 2957 2958 2959 2960
				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
2961

2962
	dp = alloc_init_deleg(oo->oo_owner.so_client, stp, fh, flag);
2963 2964
	if (dp == NULL)
		goto out_no_deleg;
2965
	status = nfs4_set_delegation(dp, flag);
2966
	if (status)
2967
		goto out_free;
L
Linus Torvalds 已提交
2968

2969
	memcpy(&open->op_delegate_stateid, &dp->dl_stid.sc_stateid, sizeof(dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2970

2971
	dprintk("NFSD: delegation stateid=" STATEID_FMT "\n",
2972
		STATEID_VAL(&dp->dl_stid.sc_stateid));
L
Linus Torvalds 已提交
2973 2974
out:
	open->op_delegate_type = flag;
2975 2976 2977 2978 2979
	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");

2980 2981 2982
		/* 4.1 client asking for a delegation? */
		if (open->op_deleg_want)
			nfsd4_open_deleg_none_ext(open, status);
2983
	}
2984 2985
	return;
out_free:
2986
	unhash_stid(&dp->dl_stid);
2987
	nfs4_put_delegation(dp);
2988 2989 2990
out_no_deleg:
	flag = NFS4_OPEN_DELEGATE_NONE;
	goto out;
L
Linus Torvalds 已提交
2991 2992
}

2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010
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
3011 3012 3013
/*
 * called with nfs4_lock_state() held.
 */
3014
__be32
L
Linus Torvalds 已提交
3015 3016
nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nfsd4_open *open)
{
A
Andy Adamson 已提交
3017
	struct nfsd4_compoundres *resp = rqstp->rq_resp;
3018
	struct nfs4_client *cl = open->op_openowner->oo_owner.so_client;
L
Linus Torvalds 已提交
3019 3020
	struct nfs4_file *fp = NULL;
	struct inode *ino = current_fh->fh_dentry->d_inode;
3021
	struct nfs4_ol_stateid *stp = NULL;
3022
	struct nfs4_delegation *dp = NULL;
3023
	__be32 status;
L
Linus Torvalds 已提交
3024 3025 3026 3027 3028 3029 3030 3031 3032 3033

	/*
	 * 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;
3034
		status = nfs4_check_deleg(cl, fp, open, &dp);
3035 3036
		if (status)
			goto out;
L
Linus Torvalds 已提交
3037
	} else {
3038
		status = nfserr_bad_stateid;
3039
		if (nfsd4_is_deleg_cur(open))
3040
			goto out;
3041
		status = nfserr_jukebox;
3042 3043 3044
		fp = open->op_file;
		open->op_file = NULL;
		nfsd4_init_file(fp, ino);
L
Linus Torvalds 已提交
3045 3046 3047 3048 3049 3050 3051 3052
	}

	/*
	 * OPEN the file, or upgrade an existing OPEN.
	 * If truncate fails, the OPEN fails.
	 */
	if (stp) {
		/* Stateid was found, this is an OPEN upgrade */
3053
		status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
L
Linus Torvalds 已提交
3054 3055 3056
		if (status)
			goto out;
	} else {
3057
		status = nfs4_get_vfs_file(rqstp, fp, current_fh, open);
J
J. Bruce Fields 已提交
3058 3059 3060
		if (status)
			goto out;
		status = nfsd4_truncate(rqstp, current_fh, open);
3061
		if (status)
L
Linus Torvalds 已提交
3062
			goto out;
3063 3064
		stp = open->op_stp;
		open->op_stp = NULL;
3065
		init_open_stateid(stp, fp, open);
L
Linus Torvalds 已提交
3066
	}
3067 3068
	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
3069

3070
	if (nfsd4_has_session(&resp->cstate)) {
3071
		open->op_openowner->oo_flags |= NFS4_OO_CONFIRMED;
A
Andy Adamson 已提交
3072

3073 3074 3075 3076 3077 3078 3079
		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
3080 3081 3082 3083
	/*
	* Attempt to hand out a delegation. No error return, because the
	* OPEN succeeds even if we fail.
	*/
3084
	nfs4_open_delegation(SVC_NET(rqstp), current_fh, open, stp);
3085
nodeleg:
L
Linus Torvalds 已提交
3086 3087
	status = nfs_ok;

3088
	dprintk("%s: stateid=" STATEID_FMT "\n", __func__,
3089
		STATEID_VAL(&stp->st_stid.sc_stateid));
L
Linus Torvalds 已提交
3090
out:
3091 3092
	/* 4.1 client trying to upgrade/downgrade delegation? */
	if (open->op_delegate_type == NFS4_OPEN_DELEGATE_NONE && dp &&
3093 3094
	    open->op_deleg_want)
		nfsd4_deleg_xgrade_none_ext(open, dp);
3095

3096 3097
	if (fp)
		put_nfs4_file(fp);
3098
	if (status == 0 && open->op_claim_type == NFS4_OPEN_CLAIM_PREVIOUS)
3099
		nfs4_set_claim_prev(open, nfsd4_has_session(&resp->cstate));
L
Linus Torvalds 已提交
3100 3101 3102 3103
	/*
	* To finish the open response, we just need to set the rflags.
	*/
	open->op_rflags = NFS4_OPEN_RESULT_LOCKTYPE_POSIX;
3104
	if (!(open->op_openowner->oo_flags & NFS4_OO_CONFIRMED) &&
A
Andy Adamson 已提交
3105
	    !nfsd4_has_session(&resp->cstate))
L
Linus Torvalds 已提交
3106 3107 3108 3109 3110
		open->op_rflags |= NFS4_OPEN_RESULT_CONFIRM;

	return status;
}

3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125
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;
		}
	}
3126 3127
	if (open->op_file)
		nfsd4_free_file(open->op_file);
3128
	if (open->op_stp)
3129
		free_generic_stateid(open->op_stp);
3130 3131
}

3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
static __be32 lookup_clientid(clientid_t *clid, bool session, struct nfsd_net *nn, struct nfs4_client **clp)
{
	struct nfs4_client *found;

	if (STALE_CLIENTID(clid, nn))
		return nfserr_stale_clientid;
	found = find_confirmed_client(clid, session, nn);
	if (clp)
		*clp = found;
	return found ? nfs_ok : nfserr_expired;
}

3144
__be32
3145 3146
nfsd4_renew(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    clientid_t *clid)
L
Linus Torvalds 已提交
3147 3148
{
	struct nfs4_client *clp;
3149
	__be32 status;
3150
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3151 3152 3153 3154

	nfs4_lock_state();
	dprintk("process_renew(%08x/%08x): starting\n", 
			clid->cl_boot, clid->cl_id);
3155 3156
	status = lookup_clientid(clid, cstate->minorversion, nn, &clp);
	if (status)
L
Linus Torvalds 已提交
3157 3158
		goto out;
	status = nfserr_cb_path_down;
3159
	if (!list_empty(&clp->cl_delegations)
3160
			&& clp->cl_cb_state != NFSD4_CB_UP)
L
Linus Torvalds 已提交
3161 3162 3163 3164 3165 3166 3167
		goto out;
	status = nfs_ok;
out:
	nfs4_unlock_state();
	return status;
}

3168
static void
3169
nfsd4_end_grace(struct nfsd_net *nn)
3170
{
3171
	/* do nothing if grace period already ended */
3172
	if (nn->grace_ended)
3173 3174
		return;

3175
	dprintk("NFSD: end of grace period\n");
3176
	nn->grace_ended = true;
3177
	nfsd4_record_grace_done(nn, nn->boot_time);
3178
	locks_end_grace(&nn->nfsd4_manager);
3179 3180 3181 3182 3183
	/*
	 * 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:
	 */
3184
	nn->nfsd4_grace = nn->nfsd4_lease;
3185 3186
}

3187
static time_t
3188
nfs4_laundromat(struct nfsd_net *nn)
L
Linus Torvalds 已提交
3189 3190
{
	struct nfs4_client *clp;
3191
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3192 3193
	struct nfs4_delegation *dp;
	struct list_head *pos, *next, reaplist;
3194 3195 3196
	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
3197 3198 3199 3200

	nfs4_lock_state();

	dprintk("NFSD: laundromat service - starting\n");
3201
	nfsd4_end_grace(nn);
3202
	INIT_LIST_HEAD(&reaplist);
3203
	spin_lock(&nn->client_lock);
3204
	list_for_each_safe(pos, next, &nn->client_lru) {
L
Linus Torvalds 已提交
3205 3206 3207 3208 3209 3210 3211
		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;
		}
3212 3213 3214 3215 3216 3217 3218
		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);
3219
	}
3220
	spin_unlock(&nn->client_lock);
3221 3222
	list_for_each_safe(pos, next, &reaplist) {
		clp = list_entry(pos, struct nfs4_client, cl_lru);
L
Linus Torvalds 已提交
3223 3224 3225 3226 3227 3228 3229
		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);
3230 3231
		if (net_generic(dp->dl_stid.sc_client->net, nfsd_net_id) != nn)
			continue;
L
Linus Torvalds 已提交
3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244
		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);
	}
3245
	test_val = nn->nfsd4_lease;
3246
	list_for_each_safe(pos, next, &nn->close_lru) {
3247 3248 3249
		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
3250 3251 3252 3253
			if (test_val > u)
				test_val = u;
			break;
		}
3254
		release_openowner(oo);
L
Linus Torvalds 已提交
3255 3256 3257 3258 3259 3260 3261
	}
	if (clientid_val < NFSD_LAUNDROMAT_MINTIMEOUT)
		clientid_val = NFSD_LAUNDROMAT_MINTIMEOUT;
	nfs4_unlock_state();
	return clientid_val;
}

H
Harvey Harrison 已提交
3262 3263 3264 3265
static struct workqueue_struct *laundry_wq;
static void laundromat_main(struct work_struct *);

static void
3266
laundromat_main(struct work_struct *laundry)
L
Linus Torvalds 已提交
3267 3268
{
	time_t t;
3269 3270 3271 3272
	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
3273

3274
	t = nfs4_laundromat(nn);
L
Linus Torvalds 已提交
3275
	dprintk("NFSD: laundromat_main - sleeping for %ld seconds\n", t);
3276
	queue_delayed_work(laundry_wq, &nn->laundromat_work, t*HZ);
L
Linus Torvalds 已提交
3277 3278
}

3279
static inline __be32 nfs4_check_fh(struct svc_fh *fhp, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3280
{
3281 3282 3283
	if (fhp->fh_dentry->d_inode != stp->st_file->fi_inode)
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3284 3285 3286
}

static int
3287
STALE_STATEID(stateid_t *stateid, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3288
{
3289
	if (stateid->si_opaque.so_clid.cl_boot == nn->boot_time)
3290 3291
		return 0;
	dprintk("NFSD: stale stateid " STATEID_FMT "!\n",
3292
		STATEID_VAL(stateid));
3293
	return 1;
L
Linus Torvalds 已提交
3294 3295 3296
}

static inline int
3297
access_permit_read(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3298
{
3299 3300 3301
	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
3302 3303 3304
}

static inline int
3305
access_permit_write(struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3306
{
3307 3308
	return test_access(NFS4_SHARE_ACCESS_WRITE, stp) ||
		test_access(NFS4_SHARE_ACCESS_BOTH, stp);
L
Linus Torvalds 已提交
3309 3310 3311
}

static
3312
__be32 nfs4_check_openmode(struct nfs4_ol_stateid *stp, int flags)
L
Linus Torvalds 已提交
3313
{
3314
        __be32 status = nfserr_openmode;
L
Linus Torvalds 已提交
3315

3316 3317 3318
	/* For lock stateid's, we test the parent open, not the lock: */
	if (stp->st_openstp)
		stp = stp->st_openstp;
3319
	if ((flags & WR_STATE) && !access_permit_write(stp))
L
Linus Torvalds 已提交
3320
                goto out;
3321
	if ((flags & RD_STATE) && !access_permit_read(stp))
L
Linus Torvalds 已提交
3322 3323 3324 3325 3326 3327
                goto out;
	status = nfs_ok;
out:
	return status;
}

3328
static inline __be32
3329
check_special_stateids(struct net *net, svc_fh *current_fh, stateid_t *stateid, int flags)
L
Linus Torvalds 已提交
3330
{
3331
	if (ONE_STATEID(stateid) && (flags & RD_STATE))
L
Linus Torvalds 已提交
3332
		return nfs_ok;
3333
	else if (locks_in_grace(net)) {
L
Lucas De Marchi 已提交
3334
		/* Answer in remaining cases depends on existence of
L
Linus Torvalds 已提交
3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349
		 * 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
3350
grace_disallows_io(struct net *net, struct inode *inode)
L
Linus Torvalds 已提交
3351
{
3352
	return locks_in_grace(net) && mandatory_lock(inode);
L
Linus Torvalds 已提交
3353 3354
}

3355 3356 3357 3358 3359 3360
/* 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
3361
static __be32 check_stateid_generation(stateid_t *in, stateid_t *ref, bool has_session)
3362
{
A
Andy Adamson 已提交
3363 3364 3365 3366
	/*
	 * When sessions are used the stateid generation number is ignored
	 * when it is zero.
	 */
J
J. Bruce Fields 已提交
3367
	if (has_session && in->si_generation == 0)
3368 3369 3370 3371
		return nfs_ok;

	if (in->si_generation == ref->si_generation)
		return nfs_ok;
A
Andy Adamson 已提交
3372

3373
	/* If the client sends us a stateid from the future, it's buggy: */
3374
	if (stateid_generation_after(in, ref))
3375 3376
		return nfserr_bad_stateid;
	/*
3377 3378 3379 3380 3381 3382 3383 3384
	 * 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:
3385
	 */
3386
	return nfserr_old_stateid;
3387 3388
}

3389
static __be32 nfsd4_validate_stateid(struct nfs4_client *cl, stateid_t *stateid)
3390
{
3391 3392 3393
	struct nfs4_stid *s;
	struct nfs4_ol_stateid *ols;
	__be32 status;
3394

3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405
	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;
	}
3406
	s = find_stateid(cl, stateid);
3407
	if (!s)
3408
		return nfserr_bad_stateid;
3409
	status = check_stateid_generation(stateid, &s->sc_stateid, 1);
3410
	if (status)
3411 3412 3413 3414 3415
		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
3416
	    && !(openowner(ols->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
3417 3418
		return nfserr_bad_stateid;
	return nfs_ok;
3419 3420
}

3421 3422 3423
static __be32 nfsd4_lookup_stateid(stateid_t *stateid, unsigned char typemask,
				   struct nfs4_stid **s, bool sessions,
				   struct nfsd_net *nn)
3424 3425 3426 3427 3428
{
	struct nfs4_client *cl;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
		return nfserr_bad_stateid;
3429
	if (STALE_STATEID(stateid, nn))
3430
		return nfserr_stale_stateid;
3431
	cl = find_confirmed_client(&stateid->si_opaque.so_clid, sessions, nn);
3432 3433 3434 3435 3436 3437 3438 3439 3440
	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
3441 3442 3443
/*
* Checks for stateid operations
*/
3444
__be32
3445
nfs4_preprocess_stateid_op(struct net *net, struct nfsd4_compound_state *cstate,
3446
			   stateid_t *stateid, int flags, struct file **filpp)
L
Linus Torvalds 已提交
3447
{
3448
	struct nfs4_stid *s;
3449
	struct nfs4_ol_stateid *stp = NULL;
L
Linus Torvalds 已提交
3450
	struct nfs4_delegation *dp = NULL;
3451
	struct svc_fh *current_fh = &cstate->current_fh;
L
Linus Torvalds 已提交
3452
	struct inode *ino = current_fh->fh_dentry->d_inode;
3453
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
3454
	__be32 status;
L
Linus Torvalds 已提交
3455 3456 3457 3458

	if (filpp)
		*filpp = NULL;

3459
	if (grace_disallows_io(net, ino))
L
Linus Torvalds 已提交
3460 3461 3462
		return nfserr_grace;

	if (ZERO_STATEID(stateid) || ONE_STATEID(stateid))
3463
		return check_special_stateids(net, current_fh, stateid, flags);
L
Linus Torvalds 已提交
3464

3465 3466
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID|NFS4_OPEN_STID|NFS4_LOCK_STID,
				      &s, cstate->minorversion, nn);
3467 3468
	if (status)
		return status;
3469 3470 3471
	status = check_stateid_generation(stateid, &s->sc_stateid, nfsd4_has_session(cstate));
	if (status)
		goto out;
3472 3473
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3474
		dp = delegstateid(s);
3475 3476 3477
		status = nfs4_check_delegmode(dp, flags);
		if (status)
			goto out;
3478
		if (filpp) {
3479
			*filpp = dp->dl_file->fi_deleg_file;
3480 3481 3482 3483 3484
			if (!*filpp) {
				WARN_ON_ONCE(1);
				status = nfserr_serverfault;
				goto out;
			}
3485
		}
3486 3487 3488
		break;
	case NFS4_OPEN_STID:
	case NFS4_LOCK_STID:
3489
		stp = openlockstateid(s);
3490 3491
		status = nfs4_check_fh(current_fh, stp);
		if (status)
L
Linus Torvalds 已提交
3492
			goto out;
3493
		if (stp->st_stateowner->so_is_open_owner
3494
		    && !(openowner(stp->st_stateowner)->oo_flags & NFS4_OO_CONFIRMED))
L
Linus Torvalds 已提交
3495
			goto out;
3496 3497
		status = nfs4_check_openmode(stp, flags);
		if (status)
L
Linus Torvalds 已提交
3498
			goto out;
3499 3500 3501 3502 3503 3504
		if (filpp) {
			if (flags & RD_STATE)
				*filpp = find_readable_file(stp->st_file);
			else
				*filpp = find_writeable_file(stp->st_file);
		}
3505 3506 3507
		break;
	default:
		return nfserr_bad_stateid;
L
Linus Torvalds 已提交
3508 3509 3510 3511 3512 3513
	}
	status = nfs_ok;
out:
	return status;
}

3514
static __be32
3515
nfsd4_free_lock_stateid(struct nfs4_ol_stateid *stp)
3516
{
3517
	if (check_for_locks(stp->st_file, lockowner(stp->st_stateowner)))
3518 3519 3520 3521 3522
		return nfserr_locks_held;
	release_lock_stateid(stp);
	return nfs_ok;
}

3523 3524 3525 3526 3527 3528 3529
/*
 * 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)
{
3530 3531 3532 3533 3534
	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)
3535 3536
		stateid->ts_id_status =
			nfsd4_validate_stateid(cl, &stateid->ts_id_stateid);
3537 3538
	nfs4_unlock_state();

3539 3540 3541
	return nfs_ok;
}

3542 3543 3544 3545 3546
__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
3547
	struct nfs4_stid *s;
3548
	struct nfs4_client *cl = cstate->session->se_client;
J
J. Bruce Fields 已提交
3549
	__be32 ret = nfserr_bad_stateid;
3550 3551

	nfs4_lock_state();
3552
	s = find_stateid(cl, stateid);
J
J. Bruce Fields 已提交
3553
	if (!s)
3554
		goto out;
J
J. Bruce Fields 已提交
3555 3556
	switch (s->sc_type) {
	case NFS4_DELEG_STID:
3557 3558
		ret = nfserr_locks_held;
		goto out;
J
J. Bruce Fields 已提交
3559 3560 3561 3562 3563 3564 3565 3566 3567
	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;
3568 3569 3570
		break;
	default:
		ret = nfserr_bad_stateid;
3571 3572 3573 3574 3575 3576
	}
out:
	nfs4_unlock_state();
	return ret;
}

3577 3578 3579 3580 3581 3582
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
3583

3584
static __be32 nfs4_seqid_op_checks(struct nfsd4_compound_state *cstate, stateid_t *stateid, u32 seqid, struct nfs4_ol_stateid *stp)
3585 3586 3587 3588 3589 3590 3591 3592
{
	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;
3593 3594 3595 3596 3597 3598 3599 3600 3601 3602
	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);
3603 3604
}

L
Linus Torvalds 已提交
3605 3606 3607
/* 
 * Checks for sequence id mutating operations. 
 */
3608
static __be32
3609
nfs4_preprocess_seqid_op(struct nfsd4_compound_state *cstate, u32 seqid,
3610
			 stateid_t *stateid, char typemask,
3611 3612
			 struct nfs4_ol_stateid **stpp,
			 struct nfsd_net *nn)
L
Linus Torvalds 已提交
3613
{
3614
	__be32 status;
3615
	struct nfs4_stid *s;
L
Linus Torvalds 已提交
3616

3617 3618
	dprintk("NFSD: %s: seqid=%d stateid = " STATEID_FMT "\n", __func__,
		seqid, STATEID_VAL(stateid));
3619

L
Linus Torvalds 已提交
3620
	*stpp = NULL;
3621 3622
	status = nfsd4_lookup_stateid(stateid, typemask, &s,
				      cstate->minorversion, nn);
3623 3624
	if (status)
		return status;
3625
	*stpp = openlockstateid(s);
3626
	cstate->replay_owner = (*stpp)->st_stateowner;
L
Linus Torvalds 已提交
3627

3628 3629
	return nfs4_seqid_op_checks(cstate, stateid, seqid, *stpp);
}
3630

3631 3632
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)
3633 3634 3635
{
	__be32 status;
	struct nfs4_openowner *oo;
L
Linus Torvalds 已提交
3636

3637
	status = nfs4_preprocess_seqid_op(cstate, seqid, stateid,
3638
						NFS4_OPEN_STID, stpp, nn);
3639 3640
	if (status)
		return status;
3641
	oo = openowner((*stpp)->st_stateowner);
3642
	if (!(oo->oo_flags & NFS4_OO_CONFIRMED))
3643 3644
		return nfserr_bad_stateid;
	return nfs_ok;
L
Linus Torvalds 已提交
3645 3646
}

3647
__be32
3648
nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3649
		   struct nfsd4_open_confirm *oc)
L
Linus Torvalds 已提交
3650
{
3651
	__be32 status;
3652
	struct nfs4_openowner *oo;
3653
	struct nfs4_ol_stateid *stp;
3654
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3655 3656

	dprintk("NFSD: nfsd4_open_confirm on file %.*s\n",
3657 3658
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3659

3660
	status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0);
3661 3662
	if (status)
		return status;
L
Linus Torvalds 已提交
3663 3664 3665

	nfs4_lock_state();

3666
	status = nfs4_preprocess_seqid_op(cstate,
3667
					oc->oc_seqid, &oc->oc_req_stateid,
3668
					NFS4_OPEN_STID, &stp, nn);
3669
	if (status)
3670
		goto out;
3671
	oo = openowner(stp->st_stateowner);
3672
	status = nfserr_bad_stateid;
3673
	if (oo->oo_flags & NFS4_OO_CONFIRMED)
3674
		goto out;
3675
	oo->oo_flags |= NFS4_OO_CONFIRMED;
3676 3677
	update_stateid(&stp->st_stid.sc_stateid);
	memcpy(&oc->oc_resp_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t));
3678
	dprintk("NFSD: %s: success, seqid=%d stateid=" STATEID_FMT "\n",
3679
		__func__, oc->oc_seqid, STATEID_VAL(&stp->st_stid.sc_stateid));
3680

3681
	nfsd4_client_record_create(oo->oo_owner.so_client);
3682
	status = nfs_ok;
L
Linus Torvalds 已提交
3683
out:
3684 3685
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3686 3687 3688
	return status;
}

J
J. Bruce Fields 已提交
3689
static inline void nfs4_stateid_downgrade_bit(struct nfs4_ol_stateid *stp, u32 access)
L
Linus Torvalds 已提交
3690
{
3691
	if (!test_access(access, stp))
J
J. Bruce Fields 已提交
3692 3693
		return;
	nfs4_file_put_access(stp->st_file, nfs4_access_to_omode(access));
3694
	clear_access(access, stp);
J
J. Bruce Fields 已提交
3695
}
3696

J
J. Bruce Fields 已提交
3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710
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:
3711
		WARN_ON_ONCE(1);
L
Linus Torvalds 已提交
3712 3713 3714 3715
	}
}

static void
3716
reset_union_bmap_deny(unsigned long deny, struct nfs4_ol_stateid *stp)
L
Linus Torvalds 已提交
3717 3718 3719 3720
{
	int i;
	for (i = 0; i < 4; i++) {
		if ((i & deny) != i)
3721
			clear_deny(i, stp);
L
Linus Torvalds 已提交
3722 3723 3724
	}
}

3725
__be32
3726 3727
nfsd4_open_downgrade(struct svc_rqst *rqstp,
		     struct nfsd4_compound_state *cstate,
3728
		     struct nfsd4_open_downgrade *od)
L
Linus Torvalds 已提交
3729
{
3730
	__be32 status;
3731
	struct nfs4_ol_stateid *stp;
3732
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3733 3734

	dprintk("NFSD: nfsd4_open_downgrade on file %.*s\n", 
3735 3736
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3737

3738
	/* We don't yet support WANT bits: */
3739 3740 3741
	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
3742 3743

	nfs4_lock_state();
3744
	status = nfs4_preprocess_confirmed_seqid_op(cstate, od->od_seqid,
3745
					&od->od_stateid, &stp, nn);
3746
	if (status)
L
Linus Torvalds 已提交
3747 3748
		goto out; 
	status = nfserr_inval;
3749 3750
	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
3751 3752 3753
			stp->st_access_bmap, od->od_share_access);
		goto out;
	}
3754
	if (!test_deny(od->od_share_deny, stp)) {
L
Linus Torvalds 已提交
3755 3756 3757 3758
		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
3759
	nfs4_stateid_downgrade(stp, od->od_share_access);
L
Linus Torvalds 已提交
3760

3761
	reset_union_bmap_deny(od->od_share_deny, stp);
L
Linus Torvalds 已提交
3762

3763 3764
	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
3765 3766
	status = nfs_ok;
out:
3767 3768
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3769 3770 3771
	return status;
}

3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788
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;
3789 3790 3791 3792 3793 3794 3795
	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;
3796 3797
}

L
Linus Torvalds 已提交
3798 3799 3800
/*
 * nfs4_unlock_state() called after encode
 */
3801
__be32
3802
nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
3803
	    struct nfsd4_close *close)
L
Linus Torvalds 已提交
3804
{
3805
	__be32 status;
3806
	struct nfs4_openowner *oo;
3807
	struct nfs4_ol_stateid *stp;
3808 3809
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
3810 3811

	dprintk("NFSD: nfsd4_close on file %.*s\n", 
3812 3813
			(int)cstate->current_fh.fh_dentry->d_name.len,
			cstate->current_fh.fh_dentry->d_name.name);
L
Linus Torvalds 已提交
3814 3815

	nfs4_lock_state();
3816 3817 3818
	status = nfs4_preprocess_seqid_op(cstate, close->cl_seqid,
					&close->cl_stateid,
					NFS4_OPEN_STID|NFS4_CLOSED_STID,
3819
					&stp, nn);
3820
	if (status)
L
Linus Torvalds 已提交
3821
		goto out; 
3822
	oo = openowner(stp->st_stateowner);
L
Linus Torvalds 已提交
3823
	status = nfs_ok;
3824 3825
	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
3826

3827
	nfsd4_close_open_stateid(stp);
J
J. Bruce Fields 已提交
3828
	release_last_closed_stateid(oo);
3829
	oo->oo_last_closed_stid = stp;
3830

3831 3832 3833 3834 3835 3836 3837 3838 3839 3840
	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))
3841
				move_to_close_lru(oo, SVC_NET(rqstp));
3842 3843
		}
	}
L
Linus Torvalds 已提交
3844
out:
3845 3846
	if (!cstate->replay_owner)
		nfs4_unlock_state();
L
Linus Torvalds 已提交
3847 3848 3849
	return status;
}

3850
__be32
3851 3852
nfsd4_delegreturn(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
		  struct nfsd4_delegreturn *dr)
L
Linus Torvalds 已提交
3853
{
3854 3855
	struct nfs4_delegation *dp;
	stateid_t *stateid = &dr->dr_stateid;
3856
	struct nfs4_stid *s;
3857
	__be32 status;
3858
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
3859

3860
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
3861
		return status;
L
Linus Torvalds 已提交
3862 3863

	nfs4_lock_state();
3864 3865
	status = nfsd4_lookup_stateid(stateid, NFS4_DELEG_STID, &s,
				      cstate->minorversion, nn);
3866
	if (status)
3867
		goto out;
3868
	dp = delegstateid(s);
3869
	status = check_stateid_generation(stateid, &dp->dl_stid.sc_stateid, nfsd4_has_session(cstate));
3870 3871 3872 3873
	if (status)
		goto out;

	unhash_delegation(dp);
L
Linus Torvalds 已提交
3874
out:
3875 3876
	nfs4_unlock_state();

L
Linus Torvalds 已提交
3877 3878 3879 3880 3881 3882
	return status;
}


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

3883
#define LOCKOWNER_INO_HASH_MASK (LOCKOWNER_INO_HASH_SIZE - 1)
L
Linus Torvalds 已提交
3884

B
Benny Halevy 已提交
3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899
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;

3900
	WARN_ON_ONCE(!len);
B
Benny Halevy 已提交
3901 3902 3903 3904
	end = start + len;
	return end > start ? end - 1: NFS4_MAX_UINT64;
}

3905
static unsigned int lockowner_ino_hashval(struct inode *inode, u32 cl_id, struct xdr_netobj *ownername)
L
Linus Torvalds 已提交
3906 3907 3908
{
	return (file_hashval(inode) + cl_id
			+ opaque_hashval(ownername->data, ownername->len))
3909
		& LOCKOWNER_INO_HASH_MASK;
L
Linus Torvalds 已提交
3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928
}

/*
 * 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;
}

3929 3930
/* 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. */
3931
static const struct lock_manager_operations nfsd_posix_mng_ops  = {
3932
};
L
Linus Torvalds 已提交
3933 3934 3935 3936

static inline void
nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
{
3937
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3938

3939
	if (fl->fl_lmops == &nfsd_posix_mng_ops) {
3940 3941 3942
		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);
3943 3944 3945
		if (!deny->ld_owner.data)
			/* We just don't care that much */
			goto nevermind;
3946 3947
		deny->ld_owner.len = lo->lo_owner.so_owner.len;
		deny->ld_clientid = lo->lo_owner.so_client->cl_clientid;
3948
	} else {
3949 3950 3951
nevermind:
		deny->ld_owner.len = 0;
		deny->ld_owner.data = NULL;
3952 3953
		deny->ld_clientid.cl_boot = 0;
		deny->ld_clientid.cl_id = 0;
L
Linus Torvalds 已提交
3954 3955
	}
	deny->ld_start = fl->fl_start;
B
Benny Halevy 已提交
3956 3957
	deny->ld_length = NFS4_MAX_UINT64;
	if (fl->fl_end != NFS4_MAX_UINT64)
L
Linus Torvalds 已提交
3958 3959 3960 3961 3962 3963
		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
3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974
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;
}

3975 3976
static struct nfs4_lockowner *
find_lockowner_str(struct inode *inode, clientid_t *clid,
3977
		   struct xdr_netobj *owner, struct nfsd_net *nn)
L
Linus Torvalds 已提交
3978
{
3979
	unsigned int hashval = lockowner_ino_hashval(inode, clid->cl_id, owner);
J
J. Bruce Fields 已提交
3980
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
3981

3982
	list_for_each_entry(lo, &nn->lockowner_ino_hashtbl[hashval], lo_owner_ino_hash) {
J
J. Bruce Fields 已提交
3983 3984
		if (same_lockowner_ino(lo, inode, clid, owner))
			return lo;
L
Linus Torvalds 已提交
3985 3986 3987 3988
	}
	return NULL;
}

3989
static void hash_lockowner(struct nfs4_lockowner *lo, unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp)
3990
{
3991 3992 3993
	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);
3994
	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
3995

3996
	list_add(&lo->lo_owner.so_strhash, &nn->ownerstr_hashtbl[strhashval]);
3997
	list_add(&lo->lo_owner_ino_hash, &nn->lockowner_ino_hashtbl[inohash]);
3998
	list_add(&lo->lo_perstateid, &open_stp->st_lockowners);
3999 4000
}

L
Linus Torvalds 已提交
4001 4002 4003
/*
 * Alloc a lock owner structure.
 * Called in nfsd4_lock - therefore, OPEN and OPEN_CONFIRM (if needed) has 
L
Lucas De Marchi 已提交
4004
 * occurred. 
L
Linus Torvalds 已提交
4005
 *
4006
 * strhashval = ownerstr_hashval
L
Linus Torvalds 已提交
4007 4008
 */

4009
static struct nfs4_lockowner *
4010
alloc_init_lock_stateowner(unsigned int strhashval, struct nfs4_client *clp, struct nfs4_ol_stateid *open_stp, struct nfsd4_lock *lock) {
4011
	struct nfs4_lockowner *lo;
L
Linus Torvalds 已提交
4012

4013 4014
	lo = alloc_stateowner(lockowner_slab, &lock->lk_new_owner, clp);
	if (!lo)
L
Linus Torvalds 已提交
4015
		return NULL;
4016 4017
	INIT_LIST_HEAD(&lo->lo_owner.so_stateids);
	lo->lo_owner.so_is_open_owner = 0;
4018 4019
	/* It is the openowner seqid that will be incremented in encode in the
	 * case of new lockowners; so increment the lock seqid manually: */
4020 4021 4022
	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
4023 4024
}

4025 4026
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
4027
{
4028
	struct nfs4_ol_stateid *stp;
4029
	struct nfs4_client *clp = lo->lo_owner.so_client;
L
Linus Torvalds 已提交
4030

4031
	stp = nfs4_alloc_stateid(clp);
N
NeilBrown 已提交
4032
	if (stp == NULL)
J
J. Bruce Fields 已提交
4033
		return NULL;
4034
	init_stid(&stp->st_stid, clp, NFS4_LOCK_STID);
4035
	list_add(&stp->st_perfile, &fp->fi_stateids);
4036 4037
	list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
	stp->st_stateowner = &lo->lo_owner;
4038
	get_nfs4_file(fp);
L
Linus Torvalds 已提交
4039
	stp->st_file = fp;
J
J. Bruce Fields 已提交
4040
	stp->st_access_bmap = 0;
L
Linus Torvalds 已提交
4041
	stp->st_deny_bmap = open_stp->st_deny_bmap;
4042
	stp->st_openstp = open_stp;
L
Linus Torvalds 已提交
4043 4044 4045
	return stp;
}

4046
static int
L
Linus Torvalds 已提交
4047 4048
check_lock_length(u64 offset, u64 length)
{
B
Benny Halevy 已提交
4049
	return ((length == 0)  || ((length != NFS4_MAX_UINT64) &&
L
Linus Torvalds 已提交
4050 4051 4052
	     LOFF_OVERFLOW(offset, length)));
}

4053
static void get_lock_access(struct nfs4_ol_stateid *lock_stp, u32 access)
J
J. Bruce Fields 已提交
4054 4055 4056 4057
{
	struct nfs4_file *fp = lock_stp->st_file;
	int oflag = nfs4_access_to_omode(access);

4058
	if (test_access(access, lock_stp))
J
J. Bruce Fields 已提交
4059 4060
		return;
	nfs4_file_get_access(fp, oflag);
4061
	set_access(access, lock_stp);
J
J. Bruce Fields 已提交
4062 4063
}

J
J. Bruce Fields 已提交
4064
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)
4065 4066 4067 4068 4069 4070
{
	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;
4071
	struct nfsd_net *nn = net_generic(cl->net, nfsd_net_id);
4072

4073 4074
	lo = find_lockowner_str(fi->fi_inode, &cl->cl_clientid,
				&lock->v.new.owner, nn);
4075 4076 4077 4078 4079 4080 4081 4082
	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;
	}
4083
	strhashval = ownerstr_hashval(cl->cl_clientid.cl_id,
4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096
			&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
4097 4098 4099
/*
 *  LOCK operation 
 */
4100
__be32
4101
nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4102
	   struct nfsd4_lock *lock)
L
Linus Torvalds 已提交
4103
{
4104 4105
	struct nfs4_openowner *open_sop = NULL;
	struct nfs4_lockowner *lock_sop = NULL;
4106
	struct nfs4_ol_stateid *lock_stp;
4107
	struct file *filp = NULL;
4108 4109
	struct file_lock *file_lock = NULL;
	struct file_lock *conflock = NULL;
4110
	__be32 status = 0;
4111
	bool new_state = false;
4112
	int lkflg;
4113
	int err;
4114 4115
	struct net *net = SVC_NET(rqstp);
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4116 4117 4118 4119 4120 4121 4122 4123

	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;

4124
	if ((status = fh_verify(rqstp, &cstate->current_fh,
M
Miklos Szeredi 已提交
4125
				S_IFREG, NFSD_MAY_LOCK))) {
A
Andy Adamson 已提交
4126 4127 4128 4129
		dprintk("NFSD: nfsd4_lock: permission denied!\n");
		return status;
	}

L
Linus Torvalds 已提交
4130 4131 4132
	nfs4_lock_state();

	if (lock->lk_is_new) {
4133
		struct nfs4_ol_stateid *open_stp = NULL;
4134 4135 4136 4137 4138 4139 4140

		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
4141
		status = nfserr_stale_clientid;
4142
		if (STALE_CLIENTID(&lock->lk_new_clientid, nn))
L
Linus Torvalds 已提交
4143 4144 4145
			goto out;

		/* validate and update open stateid and open seqid */
4146
		status = nfs4_preprocess_confirmed_seqid_op(cstate,
L
Linus Torvalds 已提交
4147 4148
				        lock->lk_new_open_seqid,
		                        &lock->lk_new_open_stateid,
4149
					&open_stp, nn);
4150
		if (status)
L
Linus Torvalds 已提交
4151
			goto out;
4152
		open_sop = openowner(open_stp->st_stateowner);
4153
		status = nfserr_bad_stateid;
4154
		if (!same_clid(&open_sop->oo_owner.so_client->cl_clientid,
4155 4156
						&lock->v.new.clientid))
			goto out;
4157 4158
		status = lookup_or_create_lock_state(cstate, open_stp, lock,
							&lock_stp, &new_state);
J
J. Bruce Fields 已提交
4159
	} else
4160
		status = nfs4_preprocess_seqid_op(cstate,
4161 4162
				       lock->lk_old_lock_seqid,
				       &lock->lk_old_lock_stateid,
4163
				       NFS4_LOCK_STID, &lock_stp, nn);
J
J. Bruce Fields 已提交
4164 4165
	if (status)
		goto out;
4166
	lock_sop = lockowner(lock_stp->st_stateowner);
L
Linus Torvalds 已提交
4167

4168 4169 4170 4171 4172
	lkflg = setlkflg(lock->lk_type);
	status = nfs4_check_openmode(lock_stp, lkflg);
	if (status)
		goto out;

4173
	status = nfserr_grace;
4174
	if (locks_in_grace(net) && !lock->lk_reclaim)
4175 4176
		goto out;
	status = nfserr_no_grace;
4177
	if (!locks_in_grace(net) && lock->lk_reclaim)
4178 4179
		goto out;

4180 4181 4182 4183 4184 4185 4186 4187
	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
4188 4189 4190
	switch (lock->lk_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
J
J. Bruce Fields 已提交
4191 4192 4193
			filp = find_readable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_READ);
4194
			file_lock->fl_type = F_RDLCK;
4195
			break;
L
Linus Torvalds 已提交
4196 4197
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
J
J. Bruce Fields 已提交
4198 4199 4200
			filp = find_writeable_file(lock_stp->st_file);
			if (filp)
				get_lock_access(lock_stp, NFS4_SHARE_ACCESS_WRITE);
4201
			file_lock->fl_type = F_WRLCK;
4202
			break;
L
Linus Torvalds 已提交
4203 4204 4205 4206
		default:
			status = nfserr_inval;
		goto out;
	}
4207 4208 4209 4210
	if (!filp) {
		status = nfserr_openmode;
		goto out;
	}
4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225
	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
4226

4227
	err = vfs_lock_file(filp, F_SETLK, file_lock, conflock);
4228
	switch (-err) {
L
Linus Torvalds 已提交
4229
	case 0: /* success! */
4230 4231
		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
4232
				sizeof(stateid_t));
4233
		status = 0;
4234 4235 4236 4237
		break;
	case (EAGAIN):		/* conflock holds conflicting lock */
		status = nfserr_denied;
		dprintk("NFSD: nfsd4_lock: conflicting lock found!\n");
4238
		nfs4_set_lock_denied(conflock, &lock->lk_denied);
4239
		break;
L
Linus Torvalds 已提交
4240 4241
	case (EDEADLK):
		status = nfserr_deadlock;
4242
		break;
4243
	default:
4244
		dprintk("NFSD: nfsd4_lock: vfs_lock_file() failed! status %d\n",err);
4245
		status = nfserrno(err);
4246
		break;
L
Linus Torvalds 已提交
4247 4248
	}
out:
4249
	if (status && new_state)
4250
		release_lockowner(lock_sop);
4251 4252
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4253 4254 4255 4256
	if (file_lock)
		locks_free_lock(file_lock);
	if (conflock)
		locks_free_lock(conflock);
L
Linus Torvalds 已提交
4257 4258 4259
	return status;
}

4260 4261 4262 4263 4264 4265
/*
 * 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.)
 */
4266
static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
4267 4268
{
	struct file *file;
4269 4270 4271 4272 4273
	__be32 err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
	if (!err) {
		err = nfserrno(vfs_test_lock(file, lock));
		nfsd_close(file);
	}
4274 4275 4276
	return err;
}

L
Linus Torvalds 已提交
4277 4278 4279
/*
 * LOCKT operation
 */
4280
__be32
4281 4282
nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
	    struct nfsd4_lockt *lockt)
L
Linus Torvalds 已提交
4283 4284
{
	struct inode *inode;
4285
	struct file_lock *file_lock = NULL;
4286
	struct nfs4_lockowner *lo;
4287
	__be32 status;
4288
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4289

4290
	if (locks_in_grace(SVC_NET(rqstp)))
L
Linus Torvalds 已提交
4291 4292 4293 4294 4295 4296 4297
		return nfserr_grace;

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

	nfs4_lock_state();

4298 4299 4300 4301 4302
	if (!nfsd4_has_session(cstate)) {
		status = lookup_clientid(&lockt->lt_clientid, false, nn, NULL);
		if (status)
			goto out;
	}
L
Linus Torvalds 已提交
4303

4304
	if ((status = fh_verify(rqstp, &cstate->current_fh, S_IFREG, 0)))
L
Linus Torvalds 已提交
4305 4306
		goto out;

4307
	inode = cstate->current_fh.fh_dentry->d_inode;
4308 4309 4310 4311 4312 4313 4314
	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
4315 4316 4317
	switch (lockt->lt_type) {
		case NFS4_READ_LT:
		case NFS4_READW_LT:
4318
			file_lock->fl_type = F_RDLCK;
L
Linus Torvalds 已提交
4319 4320 4321
		break;
		case NFS4_WRITE_LT:
		case NFS4_WRITEW_LT:
4322
			file_lock->fl_type = F_WRLCK;
L
Linus Torvalds 已提交
4323 4324
		break;
		default:
4325
			dprintk("NFSD: nfs4_lockt: bad lock type!\n");
L
Linus Torvalds 已提交
4326 4327 4328 4329
			status = nfserr_inval;
		goto out;
	}

4330
	lo = find_lockowner_str(inode, &lockt->lt_clientid, &lockt->lt_owner, nn);
4331
	if (lo)
4332 4333 4334
		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
4335

4336 4337
	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
4338

4339
	nfs4_transform_lock_offset(file_lock);
L
Linus Torvalds 已提交
4340

4341
	status = nfsd_test_lock(rqstp, &cstate->current_fh, file_lock);
4342
	if (status)
4343
		goto out;
4344

4345
	if (file_lock->fl_type != F_UNLCK) {
L
Linus Torvalds 已提交
4346
		status = nfserr_denied;
4347
		nfs4_set_lock_denied(file_lock, &lockt->lt_denied);
L
Linus Torvalds 已提交
4348 4349 4350
	}
out:
	nfs4_unlock_state();
4351 4352
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4353 4354 4355
	return status;
}

4356
__be32
4357
nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
4358
	    struct nfsd4_locku *locku)
L
Linus Torvalds 已提交
4359
{
4360
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4361
	struct file *filp = NULL;
4362
	struct file_lock *file_lock = NULL;
4363
	__be32 status;
4364
	int err;
4365 4366
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);

L
Linus Torvalds 已提交
4367 4368 4369 4370 4371 4372 4373 4374 4375
	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();
									        
4376
	status = nfs4_preprocess_seqid_op(cstate, locku->lu_seqid,
4377 4378
					&locku->lu_stateid, NFS4_LOCK_STID,
					&stp, nn);
4379
	if (status)
L
Linus Torvalds 已提交
4380
		goto out;
4381 4382 4383 4384 4385
	filp = find_any_file(stp->st_file);
	if (!filp) {
		status = nfserr_lock_range;
		goto out;
	}
4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403
	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
4404 4405 4406 4407

	/*
	*  Try to unlock the file in the VFS.
	*/
4408
	err = vfs_lock_file(filp, F_SETLK, file_lock, NULL);
4409
	if (err) {
4410
		dprintk("NFSD: nfs4_locku: vfs_lock_file failed!\n");
L
Linus Torvalds 已提交
4411 4412 4413 4414 4415
		goto out_nfserr;
	}
	/*
	* OK, unlock succeeded; the only thing left to do is update the stateid.
	*/
4416 4417
	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
4418 4419

out:
4420 4421
	if (!cstate->replay_owner)
		nfs4_unlock_state();
4422 4423
	if (file_lock)
		locks_free_lock(file_lock);
L
Linus Torvalds 已提交
4424 4425 4426
	return status;

out_nfserr:
4427
	status = nfserrno(err);
L
Linus Torvalds 已提交
4428 4429 4430 4431 4432 4433 4434 4435 4436
	goto out;
}

/*
 * returns
 * 	1: locks held by lockowner
 * 	0: no locks held by lockowner
 */
static int
4437
check_for_locks(struct nfs4_file *filp, struct nfs4_lockowner *lowner)
L
Linus Torvalds 已提交
4438 4439
{
	struct file_lock **flpp;
4440
	struct inode *inode = filp->fi_inode;
L
Linus Torvalds 已提交
4441 4442
	int status = 0;

4443
	lock_flocks();
L
Linus Torvalds 已提交
4444
	for (flpp = &inode->i_flock; *flpp != NULL; flpp = &(*flpp)->fl_next) {
4445
		if ((*flpp)->fl_owner == (fl_owner_t)lowner) {
L
Linus Torvalds 已提交
4446 4447
			status = 1;
			goto out;
4448
		}
L
Linus Torvalds 已提交
4449 4450
	}
out:
4451
	unlock_flocks();
L
Linus Torvalds 已提交
4452 4453 4454
	return status;
}

4455
__be32
4456 4457 4458
nfsd4_release_lockowner(struct svc_rqst *rqstp,
			struct nfsd4_compound_state *cstate,
			struct nfsd4_release_lockowner *rlockowner)
L
Linus Torvalds 已提交
4459 4460
{
	clientid_t *clid = &rlockowner->rl_clientid;
4461
	struct nfs4_stateowner *sop;
4462
	struct nfs4_lockowner *lo;
4463
	struct nfs4_ol_stateid *stp;
L
Linus Torvalds 已提交
4464
	struct xdr_netobj *owner = &rlockowner->rl_owner;
4465
	struct list_head matches;
4466
	unsigned int hashval = ownerstr_hashval(clid->cl_id, owner);
4467
	__be32 status;
4468
	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
L
Linus Torvalds 已提交
4469 4470 4471 4472 4473 4474

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

	nfs4_lock_state();

4475 4476 4477 4478
	status = lookup_clientid(clid, cstate->minorversion, nn, NULL);
	if (status)
		goto out;

4479 4480
	status = nfserr_locks_held;
	INIT_LIST_HEAD(&matches);
4481

4482
	list_for_each_entry(sop, &nn->ownerstr_hashtbl[hashval], so_strhash) {
4483 4484
		if (sop->so_is_open_owner)
			continue;
4485 4486 4487 4488 4489 4490 4491 4492
		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
4493
		}
4494 4495 4496 4497 4498
	}
	/* 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
4499
	while (!list_empty(&matches)) {
4500 4501
		lo = list_entry(matches.next, struct nfs4_lockowner,
								lo_list);
N
NeilBrown 已提交
4502 4503
		/* unhash_stateowner deletes so_perclient only
		 * for openowners. */
4504 4505
		list_del(&lo->lo_list);
		release_lockowner(lo);
L
Linus Torvalds 已提交
4506 4507 4508 4509 4510 4511 4512
	}
out:
	nfs4_unlock_state();
	return status;
}

static inline struct nfs4_client_reclaim *
N
NeilBrown 已提交
4513
alloc_reclaim(void)
L
Linus Torvalds 已提交
4514
{
N
NeilBrown 已提交
4515
	return kmalloc(sizeof(struct nfs4_client_reclaim), GFP_KERNEL);
L
Linus Torvalds 已提交
4516 4517
}

4518
bool
4519
nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
4520
{
4521
	struct nfs4_client_reclaim *crp;
4522

4523
	crp = nfsd4_find_reclaim_client(name, nn);
4524
	return (crp && crp->cr_clp);
4525 4526
}

L
Linus Torvalds 已提交
4527 4528 4529
/*
 * failure => all reset bets are off, nfserr_no_grace...
 */
4530
struct nfs4_client_reclaim *
4531
nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4532 4533
{
	unsigned int strhashval;
4534
	struct nfs4_client_reclaim *crp;
L
Linus Torvalds 已提交
4535

N
NeilBrown 已提交
4536 4537
	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
	crp = alloc_reclaim();
4538 4539 4540
	if (crp) {
		strhashval = clientstr_hashval(name);
		INIT_LIST_HEAD(&crp->cr_strhash);
4541
		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
4542
		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
4543
		crp->cr_clp = NULL;
4544
		nn->reclaim_str_hashtbl_size++;
4545 4546
	}
	return crp;
L
Linus Torvalds 已提交
4547 4548
}

4549
void
4550
nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
4551 4552 4553
{
	list_del(&crp->cr_strhash);
	kfree(crp);
4554
	nn->reclaim_str_hashtbl_size--;
4555 4556
}

4557
void
4558
nfs4_release_reclaim(struct nfsd_net *nn)
L
Linus Torvalds 已提交
4559 4560 4561 4562 4563
{
	struct nfs4_client_reclaim *crp = NULL;
	int i;

	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4564 4565
		while (!list_empty(&nn->reclaim_str_hashtbl[i])) {
			crp = list_entry(nn->reclaim_str_hashtbl[i].next,
L
Linus Torvalds 已提交
4566
			                struct nfs4_client_reclaim, cr_strhash);
4567
			nfs4_remove_reclaim_record(crp, nn);
L
Linus Torvalds 已提交
4568 4569
		}
	}
4570
	WARN_ON_ONCE(nn->reclaim_str_hashtbl_size);
L
Linus Torvalds 已提交
4571 4572 4573 4574
}

/*
 * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
4575
struct nfs4_client_reclaim *
4576
nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4577 4578 4579 4580
{
	unsigned int strhashval;
	struct nfs4_client_reclaim *crp = NULL;

4581
	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
L
Linus Torvalds 已提交
4582

4583
	strhashval = clientstr_hashval(recdir);
4584
	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
4585
		if (same_name(crp->cr_recdir, recdir)) {
L
Linus Torvalds 已提交
4586 4587 4588 4589 4590 4591 4592 4593 4594
			return crp;
		}
	}
	return NULL;
}

/*
* Called from OPEN. Look for clientid in reclaim list.
*/
4595
__be32
4596
nfs4_check_open_reclaim(clientid_t *clid, bool sessions, struct nfsd_net *nn)
L
Linus Torvalds 已提交
4597
{
4598 4599 4600
	struct nfs4_client *clp;

	/* find clientid in conf_id_hashtbl */
4601
	clp = find_confirmed_client(clid, sessions, nn);
4602 4603 4604 4605
	if (clp == NULL)
		return nfserr_reclaim_bad;

	return nfsd4_client_record_check(clp) ? nfserr_reclaim_bad : nfs_ok;
L
Linus Torvalds 已提交
4606 4607
}

B
Bryan Schumaker 已提交
4608 4609
#ifdef CONFIG_NFSD_FAULT_INJECTION

4610 4611 4612 4613 4614 4615
u64 nfsd_forget_client(struct nfs4_client *clp, u64 max)
{
	expire_client(clp);
	return 1;
}

4616 4617 4618
u64 nfsd_print_client(struct nfs4_client *clp, u64 num)
{
	char buf[INET6_ADDRSTRLEN];
4619
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4620 4621 4622 4623 4624 4625 4626 4627
	printk(KERN_INFO "NFS Client: %s\n", buf);
	return 1;
}

static void nfsd_print_count(struct nfs4_client *clp, unsigned int count,
			     const char *type)
{
	char buf[INET6_ADDRSTRLEN];
4628
	rpc_ntop((struct sockaddr *)&clp->cl_addr, buf, sizeof(buf));
4629 4630 4631
	printk(KERN_INFO "NFS Client: %s has %u %s\n", buf, count, type);
}

B
Bryan Schumaker 已提交
4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657
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);
}

4658 4659 4660 4661 4662 4663 4664
u64 nfsd_print_client_locks(struct nfs4_client *clp, u64 max)
{
	u64 count = nfsd_foreach_client_lock(clp, max, NULL);
	nfsd_print_count(clp, count, "locked files");
	return count;
}

4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684
static u64 nfsd_foreach_client_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);
}

4685 4686 4687 4688 4689 4690 4691
u64 nfsd_print_client_openowners(struct nfs4_client *clp, u64 max)
{
	u64 count = nfsd_foreach_client_open(clp, max, NULL);
	nfsd_print_count(clp, count, "open files");
	return count;
}

4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737
static u64 nfsd_find_all_delegations(struct nfs4_client *clp, u64 max,
				     struct list_head *victims)
{
	struct nfs4_delegation *dp, *next;
	u64 count = 0;

	list_for_each_entry_safe(dp, next, &clp->cl_delegations, dl_perclnt) {
		if (victims)
			list_move(&dp->dl_recall_lru, victims);
		if (++count == max)
			break;
	}
	return count;
}

u64 nfsd_forget_client_delegations(struct nfs4_client *clp, u64 max)
{
	struct nfs4_delegation *dp, *next;
	LIST_HEAD(victims);
	u64 count;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, &victims);
	spin_unlock(&recall_lock);

	list_for_each_entry_safe(dp, next, &victims, dl_recall_lru)
		unhash_delegation(dp);

	return count;
}

u64 nfsd_recall_client_delegations(struct nfs4_client *clp, u64 max)
{
	struct nfs4_delegation *dp, *next;
	LIST_HEAD(victims);
	u64 count;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, &victims);
	list_for_each_entry_safe(dp, next, &victims, dl_recall_lru)
		nfsd_break_one_deleg(dp);
	spin_unlock(&recall_lock);

	return count;
}

4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749
u64 nfsd_print_client_delegations(struct nfs4_client *clp, u64 max)
{
	u64 count = 0;

	spin_lock(&recall_lock);
	count = nfsd_find_all_delegations(clp, max, NULL);
	spin_unlock(&recall_lock);

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

4750
u64 nfsd_for_n_state(u64 max, u64 (*func)(struct nfs4_client *, u64))
B
Bryan Schumaker 已提交
4751 4752
{
	struct nfs4_client *clp, *next;
4753
	u64 count = 0;
4754
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);
B
Bryan Schumaker 已提交
4755

4756 4757 4758
	if (!nfsd_netns_ready(nn))
		return 0;

4759
	list_for_each_entry_safe(clp, next, &nn->client_lru, cl_lru) {
4760 4761
		count += func(clp, max - count);
		if ((max != 0) && (count >= max))
B
Bryan Schumaker 已提交
4762 4763 4764
			break;
	}

4765 4766 4767
	return count;
}

4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782
struct nfs4_client *nfsd_find_client(struct sockaddr_storage *addr, size_t addr_size)
{
	struct nfs4_client *clp;
	struct nfsd_net *nn = net_generic(current->nsproxy->net_ns, nfsd_net_id);

	if (!nfsd_netns_ready(nn))
		return NULL;

	list_for_each_entry(clp, &nn->client_lru, cl_lru) {
		if (memcmp(&clp->cl_addr, addr, addr_size) == 0)
			return clp;
	}
	return NULL;
}

B
Bryan Schumaker 已提交
4783 4784
#endif /* CONFIG_NFSD_FAULT_INJECTION */

4785
/* initialization to perform at module load time: */
L
Linus Torvalds 已提交
4786

4787
void
4788
nfs4_state_init(void)
L
Linus Torvalds 已提交
4789
{
4790
	int i;
L
Linus Torvalds 已提交
4791 4792 4793 4794 4795

	for (i = 0; i < FILE_HASH_SIZE; i++) {
		INIT_LIST_HEAD(&file_hashtbl[i]);
	}
	INIT_LIST_HEAD(&del_recall_lru);
4796 4797
}

4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818
/*
 * 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);
}

4819
static int nfs4_state_create_net(struct net *net)
4820 4821 4822 4823 4824 4825 4826
{
	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)
4827
		goto err;
4828 4829 4830 4831
	nn->unconf_id_hashtbl = kmalloc(sizeof(struct list_head) *
			CLIENT_HASH_SIZE, GFP_KERNEL);
	if (!nn->unconf_id_hashtbl)
		goto err_unconf_id;
4832 4833 4834 4835
	nn->ownerstr_hashtbl = kmalloc(sizeof(struct list_head) *
			OWNER_HASH_SIZE, GFP_KERNEL);
	if (!nn->ownerstr_hashtbl)
		goto err_ownerstr;
4836 4837 4838 4839
	nn->lockowner_ino_hashtbl = kmalloc(sizeof(struct list_head) *
			LOCKOWNER_INO_HASH_SIZE, GFP_KERNEL);
	if (!nn->lockowner_ino_hashtbl)
		goto err_lockowner_ino;
4840 4841 4842 4843
	nn->sessionid_hashtbl = kmalloc(sizeof(struct list_head) *
			SESSION_HASH_SIZE, GFP_KERNEL);
	if (!nn->sessionid_hashtbl)
		goto err_sessionid;
4844

4845
	for (i = 0; i < CLIENT_HASH_SIZE; i++) {
4846
		INIT_LIST_HEAD(&nn->conf_id_hashtbl[i]);
4847
		INIT_LIST_HEAD(&nn->unconf_id_hashtbl[i]);
4848
	}
4849 4850
	for (i = 0; i < OWNER_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->ownerstr_hashtbl[i]);
4851 4852
	for (i = 0; i < LOCKOWNER_INO_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->lockowner_ino_hashtbl[i]);
4853 4854
	for (i = 0; i < SESSION_HASH_SIZE; i++)
		INIT_LIST_HEAD(&nn->sessionid_hashtbl[i]);
4855
	nn->conf_name_tree = RB_ROOT;
4856
	nn->unconf_name_tree = RB_ROOT;
4857
	INIT_LIST_HEAD(&nn->client_lru);
4858
	INIT_LIST_HEAD(&nn->close_lru);
4859
	spin_lock_init(&nn->client_lock);
4860

4861
	INIT_DELAYED_WORK(&nn->laundromat_work, laundromat_main);
4862
	get_net(net);
4863

4864
	return 0;
4865

4866 4867
err_sessionid:
	kfree(nn->lockowner_ino_hashtbl);
4868 4869
err_lockowner_ino:
	kfree(nn->ownerstr_hashtbl);
4870 4871
err_ownerstr:
	kfree(nn->unconf_id_hashtbl);
4872 4873
err_unconf_id:
	kfree(nn->conf_id_hashtbl);
4874 4875
err:
	return -ENOMEM;
4876 4877 4878
}

static void
4879
nfs4_state_destroy_net(struct net *net)
4880 4881 4882 4883
{
	int i;
	struct nfs4_client *clp = NULL;
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4884
	struct rb_node *node, *tmp;
4885 4886 4887 4888 4889 4890 4891

	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);
		}
	}
4892 4893 4894 4895 4896 4897 4898 4899 4900 4901

	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);
	}

4902
	kfree(nn->sessionid_hashtbl);
4903
	kfree(nn->lockowner_ino_hashtbl);
4904
	kfree(nn->ownerstr_hashtbl);
4905
	kfree(nn->unconf_id_hashtbl);
4906
	kfree(nn->conf_id_hashtbl);
4907
	put_net(net);
4908 4909
}

4910
int
4911
nfs4_state_start_net(struct net *net)
4912
{
4913
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
4914 4915
	int ret;

4916 4917 4918 4919 4920 4921 4922 4923 4924 4925
	/*
	 * 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;

4926
	ret = nfs4_state_create_net(net);
4927 4928
	if (ret)
		return ret;
4929
	nfsd4_client_tracking_init(net);
4930
	nn->boot_time = get_seconds();
4931
	locks_start_grace(net, &nn->nfsd4_manager);
4932
	nn->grace_ended = false;
4933
	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %p)\n",
4934 4935
	       nn->nfsd4_grace, net);
	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
4936 4937 4938 4939 4940 4941 4942 4943 4944 4945
	return 0;
}

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

int
nfs4_state_start(void)
{
	int ret;

4946
	ret = set_callback_cred();
4947 4948
	if (ret)
		return -ENOMEM;
4949
	laundry_wq = create_singlethread_workqueue("nfsd4");
4950 4951 4952 4953
	if (laundry_wq == NULL) {
		ret = -ENOMEM;
		goto out_recovery;
	}
4954 4955 4956
	ret = nfsd4_create_callback_queue();
	if (ret)
		goto out_free_laundry;
4957

4958
	set_max_delegations();
4959

4960
	return 0;
4961

4962 4963
out_free_laundry:
	destroy_workqueue(laundry_wq);
4964
out_recovery:
4965
	return ret;
L
Linus Torvalds 已提交
4966 4967
}

4968
/* should be called with the state lock held */
4969
void
4970
nfs4_state_shutdown_net(struct net *net)
L
Linus Torvalds 已提交
4971 4972 4973
{
	struct nfs4_delegation *dp = NULL;
	struct list_head *pos, *next, reaplist;
4974
	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
L
Linus Torvalds 已提交
4975

4976 4977
	cancel_delayed_work_sync(&nn->laundromat_work);
	locks_end_grace(&nn->nfsd4_manager);
4978

L
Linus Torvalds 已提交
4979 4980 4981 4982
	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);
4983 4984
		if (dp->dl_stid.sc_client->net != net)
			continue;
L
Linus Torvalds 已提交
4985 4986 4987 4988 4989 4990 4991 4992
		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);
	}

4993
	nfsd4_client_tracking_exit(net);
4994
	nfs4_state_destroy_net(net);
L
Linus Torvalds 已提交
4995 4996 4997 4998 4999
}

void
nfs4_state_shutdown(void)
{
5000
	destroy_workqueue(laundry_wq);
5001
	nfsd4_destroy_callback_queue();
L
Linus Torvalds 已提交
5002
}
5003 5004 5005 5006

static void
get_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5007 5008
	if (HAS_STATE_ID(cstate, CURRENT_STATE_ID_FLAG) && CURRENT_STATEID(stateid))
		memcpy(stateid, &cstate->current_stateid, sizeof(stateid_t));
5009 5010 5011 5012 5013
}

static void
put_stateid(struct nfsd4_compound_state *cstate, stateid_t *stateid)
{
5014 5015 5016 5017 5018 5019 5020 5021 5022 5023
	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);
5024 5025
}

5026 5027 5028
/*
 * functions to set current state id
 */
5029 5030 5031 5032 5033 5034
void
nfsd4_set_opendowngradestateid(struct nfsd4_compound_state *cstate, struct nfsd4_open_downgrade *odp)
{
	put_stateid(cstate, &odp->od_stateid);
}

5035 5036 5037 5038 5039 5040
void
nfsd4_set_openstateid(struct nfsd4_compound_state *cstate, struct nfsd4_open *open)
{
	put_stateid(cstate, &open->op_stateid);
}

5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055
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
 */
5056

5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068
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);
}

5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080
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);
}

5081 5082 5083 5084 5085 5086 5087
void
nfsd4_get_closestateid(struct nfsd4_compound_state *cstate, struct nfsd4_close *close)
{
	get_stateid(cstate, &close->cl_stateid);
}

void
5088
nfsd4_get_lockustateid(struct nfsd4_compound_state *cstate, struct nfsd4_locku *locku)
5089
{
5090
	get_stateid(cstate, &locku->lu_stateid);
5091
}
5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103

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);
}