nfs4state.c 21.0 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 35 36 37 38 39 40 41 42 43 44
/*
 *  fs/nfs/nfs4state.c
 *
 *  Client-side XDR for NFSv4.
 *
 *  Copyright (c) 2002 The Regents of the University of Michigan.
 *  All rights reserved.
 *
 *  Kendrick Smith <kmsmith@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.
 *
 * Implementation of the NFSv4 state model.  For the time being,
 * this is minimal, but will be made much more complex in a
 * subsequent patch.
 */

#include <linux/slab.h>
#include <linux/smp_lock.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_idmap.h>
45 46
#include <linux/kthread.h>
#include <linux/module.h>
L
Linus Torvalds 已提交
47 48 49
#include <linux/workqueue.h>
#include <linux/bitops.h>

50
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
51 52
#include "callback.h"
#include "delegation.h"
53
#include "internal.h"
L
Linus Torvalds 已提交
54 55 56

#define OPENOWNER_POOL_SIZE	8

57
const nfs4_stateid zero_stateid;
L
Linus Torvalds 已提交
58 59 60

static LIST_HEAD(nfs4_clientid_list);

61
static int nfs4_init_client(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
62
{
63 64
	int status = nfs4_proc_setclientid(clp, NFS4_CALLBACK,
			nfs_callback_tcpport, cred);
L
Linus Torvalds 已提交
65
	if (status == 0)
66
		status = nfs4_proc_setclientid_confirm(clp, cred);
L
Linus Torvalds 已提交
67 68 69 70 71 72
	if (status == 0)
		nfs4_schedule_state_renewal(clp);
	return status;
}

u32
73
nfs4_alloc_lockowner_id(struct nfs_client *clp)
L
Linus Torvalds 已提交
74 75 76 77
{
	return clp->cl_lockowner_id ++;
}

78
struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
79 80 81 82 83 84 85 86 87 88 89 90 91
{
	struct nfs4_state_owner *sp;
	struct rpc_cred *cred = NULL;

	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
		if (list_empty(&sp->so_states))
			continue;
		cred = get_rpccred(sp->so_cred);
		break;
	}
	return cred;
}

92
static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
93 94 95 96 97 98 99 100 101 102 103
{
	struct nfs4_state_owner *sp;

	if (!list_empty(&clp->cl_state_owners)) {
		sp = list_entry(clp->cl_state_owners.next,
				struct nfs4_state_owner, so_list);
		return get_rpccred(sp->so_cred);
	}
	return NULL;
}

L
Linus Torvalds 已提交
104
static struct nfs4_state_owner *
105
nfs4_find_state_owner(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
{
	struct nfs4_state_owner *sp, *res = NULL;

	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
		if (sp->so_cred != cred)
			continue;
		atomic_inc(&sp->so_count);
		/* Move to the head of the list */
		list_move(&sp->so_list, &clp->cl_state_owners);
		res = sp;
		break;
	}
	return res;
}

/*
 * nfs4_alloc_state_owner(): this is called on the OPEN or CREATE path to
 * create a new state_owner.
 *
 */
static struct nfs4_state_owner *
nfs4_alloc_state_owner(void)
{
	struct nfs4_state_owner *sp;

131
	sp = kzalloc(sizeof(*sp),GFP_KERNEL);
L
Linus Torvalds 已提交
132 133
	if (!sp)
		return NULL;
134
	spin_lock_init(&sp->so_lock);
L
Linus Torvalds 已提交
135 136
	INIT_LIST_HEAD(&sp->so_states);
	INIT_LIST_HEAD(&sp->so_delegations);
137 138 139 140
	rpc_init_wait_queue(&sp->so_sequence.wait, "Seqid_waitqueue");
	sp->so_seqid.sequence = &sp->so_sequence;
	spin_lock_init(&sp->so_sequence.lock);
	INIT_LIST_HEAD(&sp->so_sequence.list);
L
Linus Torvalds 已提交
141 142 143 144 145 146 147
	atomic_set(&sp->so_count, 1);
	return sp;
}

void
nfs4_drop_state_owner(struct nfs4_state_owner *sp)
{
148
	struct nfs_client *clp = sp->so_client;
L
Linus Torvalds 已提交
149 150 151 152 153 154 155 156 157 158 159
	spin_lock(&clp->cl_lock);
	list_del_init(&sp->so_list);
	spin_unlock(&clp->cl_lock);
}

/*
 * Note: must be called with clp->cl_sem held in order to prevent races
 *       with reboot recovery!
 */
struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
{
160
	struct nfs_client *clp = server->nfs_client;
L
Linus Torvalds 已提交
161 162 163 164 165 166 167 168 169
	struct nfs4_state_owner *sp, *new;

	new = nfs4_alloc_state_owner();
	spin_lock(&clp->cl_lock);
	sp = nfs4_find_state_owner(clp, cred);
	if (sp == NULL && new != NULL) {
		list_add(&new->so_list, &clp->cl_state_owners);
		new->so_client = clp;
		new->so_id = nfs4_alloc_lockowner_id(clp);
170
		new->so_cred = get_rpccred(cred);
L
Linus Torvalds 已提交
171 172 173 174
		sp = new;
		new = NULL;
	}
	spin_unlock(&clp->cl_lock);
J
Jesper Juhl 已提交
175
	kfree(new);
L
Linus Torvalds 已提交
176 177 178 179 180 181 182 183 184 185 186
	if (sp != NULL)
		return sp;
	return NULL;
}

/*
 * Must be called with clp->cl_sem held in order to avoid races
 * with state recovery...
 */
void nfs4_put_state_owner(struct nfs4_state_owner *sp)
{
187
	struct nfs_client *clp = sp->so_client;
L
Linus Torvalds 已提交
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202
	struct rpc_cred *cred = sp->so_cred;

	if (!atomic_dec_and_lock(&sp->so_count, &clp->cl_lock))
		return;
	list_del(&sp->so_list);
	spin_unlock(&clp->cl_lock);
	put_rpccred(cred);
	kfree(sp);
}

static struct nfs4_state *
nfs4_alloc_open_state(void)
{
	struct nfs4_state *state;

203
	state = kzalloc(sizeof(*state), GFP_KERNEL);
L
Linus Torvalds 已提交
204 205 206 207
	if (!state)
		return NULL;
	atomic_set(&state->count, 1);
	INIT_LIST_HEAD(&state->lock_states);
208
	spin_lock_init(&state->state_lock);
L
Linus Torvalds 已提交
209 210 211
	return state;
}

212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
void
nfs4_state_set_mode_locked(struct nfs4_state *state, mode_t mode)
{
	if (state->state == mode)
		return;
	/* NB! List reordering - see the reclaim code for why.  */
	if ((mode & FMODE_WRITE) != (state->state & FMODE_WRITE)) {
		if (mode & FMODE_WRITE)
			list_move(&state->open_states, &state->owner->so_states);
		else
			list_move_tail(&state->open_states, &state->owner->so_states);
	}
	if (mode == 0)
		list_del_init(&state->inode_states);
	state->state = mode;
}

L
Linus Torvalds 已提交
229 230 231 232 233 234 235 236
static struct nfs4_state *
__nfs4_find_state_byowner(struct inode *inode, struct nfs4_state_owner *owner)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs4_state *state;

	list_for_each_entry(state, &nfsi->open_states, inode_states) {
		/* Is this in the process of being freed? */
237
		if (state->state == 0)
L
Linus Torvalds 已提交
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
			continue;
		if (state->owner == owner) {
			atomic_inc(&state->count);
			return state;
		}
	}
	return NULL;
}

static void
nfs4_free_open_state(struct nfs4_state *state)
{
	kfree(state);
}

struct nfs4_state *
nfs4_get_open_state(struct inode *inode, struct nfs4_state_owner *owner)
{
	struct nfs4_state *state, *new;
	struct nfs_inode *nfsi = NFS_I(inode);

	spin_lock(&inode->i_lock);
	state = __nfs4_find_state_byowner(inode, owner);
	spin_unlock(&inode->i_lock);
	if (state)
		goto out;
	new = nfs4_alloc_open_state();
265
	spin_lock(&owner->so_lock);
L
Linus Torvalds 已提交
266 267 268 269 270 271 272 273 274
	spin_lock(&inode->i_lock);
	state = __nfs4_find_state_byowner(inode, owner);
	if (state == NULL && new != NULL) {
		state = new;
		state->owner = owner;
		atomic_inc(&owner->so_count);
		list_add(&state->inode_states, &nfsi->open_states);
		state->inode = igrab(inode);
		spin_unlock(&inode->i_lock);
275 276 277 278
		/* Note: The reclaim code dictates that we add stateless
		 * and read-only stateids to the end of the list */
		list_add_tail(&state->open_states, &owner->so_states);
		spin_unlock(&owner->so_lock);
L
Linus Torvalds 已提交
279 280
	} else {
		spin_unlock(&inode->i_lock);
281
		spin_unlock(&owner->so_lock);
L
Linus Torvalds 已提交
282 283 284 285 286 287 288 289 290
		if (new)
			nfs4_free_open_state(new);
	}
out:
	return state;
}

/*
 * Beware! Caller must be holding exactly one
291
 * reference to clp->cl_sem!
L
Linus Torvalds 已提交
292 293 294 295 296 297
 */
void nfs4_put_open_state(struct nfs4_state *state)
{
	struct inode *inode = state->inode;
	struct nfs4_state_owner *owner = state->owner;

298
	if (!atomic_dec_and_lock(&state->count, &owner->so_lock))
L
Linus Torvalds 已提交
299
		return;
300
	spin_lock(&inode->i_lock);
L
Linus Torvalds 已提交
301 302 303
	if (!list_empty(&state->inode_states))
		list_del(&state->inode_states);
	list_del(&state->open_states);
304 305
	spin_unlock(&inode->i_lock);
	spin_unlock(&owner->so_lock);
L
Linus Torvalds 已提交
306 307 308 309 310 311
	iput(inode);
	nfs4_free_open_state(state);
	nfs4_put_state_owner(owner);
}

/*
312
 * Close the current file.
L
Linus Torvalds 已提交
313
 */
314
void nfs4_close_state(struct path *path, struct nfs4_state *state, mode_t mode)
L
Linus Torvalds 已提交
315 316 317
{
	struct inode *inode = state->inode;
	struct nfs4_state_owner *owner = state->owner;
318
	int oldstate, newstate = 0;
L
Linus Torvalds 已提交
319 320 321

	atomic_inc(&owner->so_count);
	/* Protect against nfs4_find_state() */
322
	spin_lock(&owner->so_lock);
L
Linus Torvalds 已提交
323
	spin_lock(&inode->i_lock);
324 325 326 327 328 329 330 331 332 333
	switch (mode & (FMODE_READ | FMODE_WRITE)) {
		case FMODE_READ:
			state->n_rdonly--;
			break;
		case FMODE_WRITE:
			state->n_wronly--;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr--;
	}
334
	oldstate = newstate = state->state;
335 336 337 338 339 340
	if (state->n_rdwr == 0) {
		if (state->n_rdonly == 0)
			newstate &= ~FMODE_READ;
		if (state->n_wronly == 0)
			newstate &= ~FMODE_WRITE;
	}
341 342 343
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		nfs4_state_set_mode_locked(state, newstate);
		oldstate = newstate;
L
Linus Torvalds 已提交
344 345
	}
	spin_unlock(&inode->i_lock);
346
	spin_unlock(&owner->so_lock);
347

348 349 350 351 352
	if (oldstate == newstate) {
		nfs4_put_open_state(state);
		nfs4_put_state_owner(owner);
	} else
		nfs4_do_close(path, state);
L
Linus Torvalds 已提交
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379
}

/*
 * Search the state->lock_states for an existing lock_owner
 * that is compatible with current->files
 */
static struct nfs4_lock_state *
__nfs4_find_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
{
	struct nfs4_lock_state *pos;
	list_for_each_entry(pos, &state->lock_states, ls_locks) {
		if (pos->ls_owner != fl_owner)
			continue;
		atomic_inc(&pos->ls_count);
		return pos;
	}
	return NULL;
}

/*
 * Return a compatible lock_state. If no initialized lock_state structure
 * exists, return an uninitialized one.
 *
 */
static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
{
	struct nfs4_lock_state *lsp;
380
	struct nfs_client *clp = state->owner->so_client;
L
Linus Torvalds 已提交
381

382
	lsp = kzalloc(sizeof(*lsp), GFP_KERNEL);
L
Linus Torvalds 已提交
383 384
	if (lsp == NULL)
		return NULL;
385
	lsp->ls_seqid.sequence = &state->owner->so_sequence;
L
Linus Torvalds 已提交
386 387 388 389 390
	atomic_set(&lsp->ls_count, 1);
	lsp->ls_owner = fl_owner;
	spin_lock(&clp->cl_lock);
	lsp->ls_id = nfs4_alloc_lockowner_id(clp);
	spin_unlock(&clp->cl_lock);
391
	INIT_LIST_HEAD(&lsp->ls_locks);
L
Linus Torvalds 已提交
392 393 394 395 396 397 398
	return lsp;
}

/*
 * Return a compatible lock_state. If no initialized lock_state structure
 * exists, return an uninitialized one.
 *
399
 * The caller must be holding clp->cl_sem
L
Linus Torvalds 已提交
400
 */
401
static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
L
Linus Torvalds 已提交
402
{
403
	struct nfs4_lock_state *lsp, *new = NULL;
L
Linus Torvalds 已提交
404
	
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
	for(;;) {
		spin_lock(&state->state_lock);
		lsp = __nfs4_find_lock_state(state, owner);
		if (lsp != NULL)
			break;
		if (new != NULL) {
			new->ls_state = state;
			list_add(&new->ls_locks, &state->lock_states);
			set_bit(LK_STATE_IN_USE, &state->flags);
			lsp = new;
			new = NULL;
			break;
		}
		spin_unlock(&state->state_lock);
		new = nfs4_alloc_lock_state(state, owner);
		if (new == NULL)
			return NULL;
	}
	spin_unlock(&state->state_lock);
	kfree(new);
L
Linus Torvalds 已提交
425 426 427 428
	return lsp;
}

/*
429 430
 * Release reference to lock_state, and free it if we see that
 * it is no longer in use
L
Linus Torvalds 已提交
431
 */
432
void nfs4_put_lock_state(struct nfs4_lock_state *lsp)
L
Linus Torvalds 已提交
433
{
434
	struct nfs4_state *state;
L
Linus Torvalds 已提交
435

436 437 438 439 440 441 442 443 444 445
	if (lsp == NULL)
		return;
	state = lsp->ls_state;
	if (!atomic_dec_and_lock(&lsp->ls_count, &state->state_lock))
		return;
	list_del(&lsp->ls_locks);
	if (list_empty(&state->lock_states))
		clear_bit(LK_STATE_IN_USE, &state->flags);
	spin_unlock(&state->state_lock);
	kfree(lsp);
L
Linus Torvalds 已提交
446 447
}

448
static void nfs4_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
L
Linus Torvalds 已提交
449
{
450
	struct nfs4_lock_state *lsp = src->fl_u.nfs4_fl.owner;
L
Linus Torvalds 已提交
451

452 453 454
	dst->fl_u.nfs4_fl.owner = lsp;
	atomic_inc(&lsp->ls_count);
}
L
Linus Torvalds 已提交
455

456
static void nfs4_fl_release_lock(struct file_lock *fl)
L
Linus Torvalds 已提交
457
{
458
	nfs4_put_lock_state(fl->fl_u.nfs4_fl.owner);
L
Linus Torvalds 已提交
459 460
}

461 462 463 464 465 466
static struct file_lock_operations nfs4_fl_lock_ops = {
	.fl_copy_lock = nfs4_fl_copy_lock,
	.fl_release_private = nfs4_fl_release_lock,
};

int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl)
L
Linus Torvalds 已提交
467
{
468 469 470 471 472 473 474 475 476 477
	struct nfs4_lock_state *lsp;

	if (fl->fl_ops != NULL)
		return 0;
	lsp = nfs4_get_lock_state(state, fl->fl_owner);
	if (lsp == NULL)
		return -ENOMEM;
	fl->fl_u.nfs4_fl.owner = lsp;
	fl->fl_ops = &nfs4_fl_lock_ops;
	return 0;
L
Linus Torvalds 已提交
478 479
}

480 481 482
/*
 * Byte-range lock aware utility to initialize the stateid of read/write
 * requests.
L
Linus Torvalds 已提交
483
 */
484
void nfs4_copy_stateid(nfs4_stateid *dst, struct nfs4_state *state, fl_owner_t fl_owner)
L
Linus Torvalds 已提交
485
{
486
	struct nfs4_lock_state *lsp;
L
Linus Torvalds 已提交
487

488 489 490
	memcpy(dst, &state->stateid, sizeof(*dst));
	if (test_bit(LK_STATE_IN_USE, &state->flags) == 0)
		return;
L
Linus Torvalds 已提交
491

492 493 494 495 496
	spin_lock(&state->state_lock);
	lsp = __nfs4_find_lock_state(state, fl_owner);
	if (lsp != NULL && (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0)
		memcpy(dst, &lsp->ls_stateid, sizeof(*dst));
	spin_unlock(&state->state_lock);
L
Linus Torvalds 已提交
497 498 499
	nfs4_put_lock_state(lsp);
}

500 501
struct nfs_seqid *nfs_alloc_seqid(struct nfs_seqid_counter *counter)
{
502
	struct rpc_sequence *sequence = counter->sequence;
503 504 505 506 507
	struct nfs_seqid *new;

	new = kmalloc(sizeof(*new), GFP_KERNEL);
	if (new != NULL) {
		new->sequence = counter;
508 509 510
		spin_lock(&sequence->lock);
		list_add_tail(&new->list, &sequence->list);
		spin_unlock(&sequence->lock);
511 512 513 514 515
	}
	return new;
}

void nfs_free_seqid(struct nfs_seqid *seqid)
L
Linus Torvalds 已提交
516
{
517 518
	struct rpc_sequence *sequence = seqid->sequence->sequence;

519 520 521 522
	spin_lock(&sequence->lock);
	list_del(&seqid->list);
	spin_unlock(&sequence->lock);
	rpc_wake_up(&sequence->wait);
523
	kfree(seqid);
L
Linus Torvalds 已提交
524 525 526
}

/*
527 528 529 530
 * Increment the seqid if the OPEN/OPEN_DOWNGRADE/CLOSE succeeded, or
 * failed with a seqid incrementing error -
 * see comments nfs_fs.h:seqid_mutating_error()
 */
531
static void nfs_increment_seqid(int status, struct nfs_seqid *seqid)
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
{
	switch (status) {
		case 0:
			break;
		case -NFS4ERR_BAD_SEQID:
		case -NFS4ERR_STALE_CLIENTID:
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_BADXDR:
		case -NFS4ERR_RESOURCE:
		case -NFS4ERR_NOFILEHANDLE:
			/* Non-seqid mutating errors */
			return;
	};
	/*
	 * Note: no locking needed as we are guaranteed to be first
	 * on the sequence list
	 */
	seqid->sequence->counter++;
}

void nfs_increment_open_seqid(int status, struct nfs_seqid *seqid)
{
	if (status == -NFS4ERR_BAD_SEQID) {
		struct nfs4_state_owner *sp = container_of(seqid->sequence,
				struct nfs4_state_owner, so_seqid);
L
Linus Torvalds 已提交
558
		nfs4_drop_state_owner(sp);
559
	}
560
	nfs_increment_seqid(status, seqid);
561 562 563 564 565 566 567 568 569
}

/*
 * Increment the seqid if the LOCK/LOCKU succeeded, or
 * failed with a seqid incrementing error -
 * see comments nfs_fs.h:seqid_mutating_error()
 */
void nfs_increment_lock_seqid(int status, struct nfs_seqid *seqid)
{
570
	nfs_increment_seqid(status, seqid);
571 572 573 574 575 576 577
}

int nfs_wait_on_sequence(struct nfs_seqid *seqid, struct rpc_task *task)
{
	struct rpc_sequence *sequence = seqid->sequence->sequence;
	int status = 0;

578 579
	if (sequence->list.next == &seqid->list)
		goto out;
580
	spin_lock(&sequence->lock);
581
	if (sequence->list.next != &seqid->list) {
582 583
		rpc_sleep_on(&sequence->wait, task, NULL, NULL);
		status = -EAGAIN;
584
	}
585
	spin_unlock(&sequence->lock);
586
out:
587
	return status;
L
Linus Torvalds 已提交
588 589 590 591
}

static int reclaimer(void *);

592
static inline void nfs4_clear_recover_bit(struct nfs_client *clp)
T
Trond Myklebust 已提交
593 594 595 596 597 598 599 600
{
	smp_mb__before_clear_bit();
	clear_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state);
	smp_mb__after_clear_bit();
	wake_up_bit(&clp->cl_state, NFS4CLNT_STATE_RECOVER);
	rpc_wake_up(&clp->cl_rpcwaitq);
}

L
Linus Torvalds 已提交
601 602 603
/*
 * State recovery routine
 */
604
static void nfs4_recover_state(struct nfs_client *clp)
L
Linus Torvalds 已提交
605
{
606
	struct task_struct *task;
L
Linus Torvalds 已提交
607

608 609 610
	__module_get(THIS_MODULE);
	atomic_inc(&clp->cl_count);
	task = kthread_run(reclaimer, clp, "%u.%u.%u.%u-reclaim",
611
			NIPQUAD(clp->cl_addr.sin_addr));
612 613
	if (!IS_ERR(task))
		return;
T
Trond Myklebust 已提交
614
	nfs4_clear_recover_bit(clp);
615
	nfs_put_client(clp);
616
	module_put(THIS_MODULE);
L
Linus Torvalds 已提交
617 618 619 620 621
}

/*
 * Schedule a state recovery attempt
 */
622
void nfs4_schedule_state_recovery(struct nfs_client *clp)
L
Linus Torvalds 已提交
623 624 625
{
	if (!clp)
		return;
T
Trond Myklebust 已提交
626
	if (test_and_set_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
627
		nfs4_recover_state(clp);
L
Linus Torvalds 已提交
628 629 630 631 632 633 634 635 636
}

static int nfs4_reclaim_locks(struct nfs4_state_recovery_ops *ops, struct nfs4_state *state)
{
	struct inode *inode = state->inode;
	struct file_lock *fl;
	int status = 0;

	for (fl = inode->i_flock; fl != 0; fl = fl->fl_next) {
637
		if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
L
Linus Torvalds 已提交
638 639 640 641 642 643 644 645 646 647 648 649 650 651
			continue;
		if (((struct nfs_open_context *)fl->fl_file->private_data)->state != state)
			continue;
		status = ops->recover_lock(state, fl);
		if (status >= 0)
			continue;
		switch (status) {
			default:
				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
						__FUNCTION__, status);
			case -NFS4ERR_EXPIRED:
			case -NFS4ERR_NO_GRACE:
			case -NFS4ERR_RECLAIM_BAD:
			case -NFS4ERR_RECLAIM_CONFLICT:
652
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
L
Linus Torvalds 已提交
653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718
				break;
			case -NFS4ERR_STALE_CLIENTID:
				goto out_err;
		}
	}
	return 0;
out_err:
	return status;
}

static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct nfs4_state_owner *sp)
{
	struct nfs4_state *state;
	struct nfs4_lock_state *lock;
	int status = 0;

	/* Note: we rely on the sp->so_states list being ordered 
	 * so that we always reclaim open(O_RDWR) and/or open(O_WRITE)
	 * states first.
	 * This is needed to ensure that the server won't give us any
	 * read delegations that we have to return if, say, we are
	 * recovering after a network partition or a reboot from a
	 * server that doesn't support a grace period.
	 */
	list_for_each_entry(state, &sp->so_states, open_states) {
		if (state->state == 0)
			continue;
		status = ops->recover_open(sp, state);
		if (status >= 0) {
			status = nfs4_reclaim_locks(ops, state);
			if (status < 0)
				goto out_err;
			list_for_each_entry(lock, &state->lock_states, ls_locks) {
				if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
					printk("%s: Lock reclaim failed!\n",
							__FUNCTION__);
			}
			continue;
		}
		switch (status) {
			default:
				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
						__FUNCTION__, status);
			case -ENOENT:
			case -NFS4ERR_RECLAIM_BAD:
			case -NFS4ERR_RECLAIM_CONFLICT:
				/*
				 * Open state on this file cannot be recovered
				 * All we can do is revert to using the zero stateid.
				 */
				memset(state->stateid.data, 0,
					sizeof(state->stateid.data));
				/* Mark the file as being 'closed' */
				state->state = 0;
				break;
			case -NFS4ERR_EXPIRED:
			case -NFS4ERR_NO_GRACE:
			case -NFS4ERR_STALE_CLIENTID:
				goto out_err;
		}
	}
	return 0;
out_err:
	return status;
}

719
static void nfs4_state_mark_reclaim(struct nfs_client *clp)
720 721 722 723 724 725 726 727 728
{
	struct nfs4_state_owner *sp;
	struct nfs4_state *state;
	struct nfs4_lock_state *lock;

	/* Reset all sequence ids to zero */
	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
		sp->so_seqid.counter = 0;
		sp->so_seqid.flags = 0;
729
		spin_lock(&sp->so_lock);
730 731 732 733 734 735 736
		list_for_each_entry(state, &sp->so_states, open_states) {
			list_for_each_entry(lock, &state->lock_states, ls_locks) {
				lock->ls_seqid.counter = 0;
				lock->ls_seqid.flags = 0;
				lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
			}
		}
737
		spin_unlock(&sp->so_lock);
738 739 740
	}
}

L
Linus Torvalds 已提交
741 742
static int reclaimer(void *ptr)
{
743
	struct nfs_client *clp = ptr;
L
Linus Torvalds 已提交
744 745
	struct nfs4_state_owner *sp;
	struct nfs4_state_recovery_ops *ops;
746
	struct rpc_cred *cred;
L
Linus Torvalds 已提交
747 748 749 750 751 752 753 754 755 756 757
	int status = 0;

	allow_signal(SIGKILL);

	/* Ensure exclusive access to NFSv4 state */
	lock_kernel();
	down_write(&clp->cl_sem);
	/* Are there any NFS mounts out there? */
	if (list_empty(&clp->cl_superblocks))
		goto out;
restart_loop:
758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
	ops = &nfs4_network_partition_recovery_ops;
	/* Are there any open files on this volume? */
	cred = nfs4_get_renew_cred(clp);
	if (cred != NULL) {
		/* Yes there are: try to renew the old lease */
		status = nfs4_proc_renew(clp, cred);
		switch (status) {
			case 0:
			case -NFS4ERR_CB_PATH_DOWN:
				put_rpccred(cred);
				goto out;
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_LEASE_MOVED:
				ops = &nfs4_reboot_recovery_ops;
		}
	} else {
		/* "reboot" to ensure we clear all state on the server */
		clp->cl_boot_time = CURRENT_TIME;
		cred = nfs4_get_setclientid_cred(clp);
	}
	/* We're going to have to re-establish a clientid */
779
	nfs4_state_mark_reclaim(clp);
780 781 782 783 784
	status = -ENOENT;
	if (cred != NULL) {
		status = nfs4_init_client(clp, cred);
		put_rpccred(cred);
	}
L
Linus Torvalds 已提交
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808
	if (status)
		goto out_error;
	/* Mark all delegations for reclaim */
	nfs_delegation_mark_reclaim(clp);
	/* Note: list is protected by exclusive lock on cl->cl_sem */
	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
		status = nfs4_reclaim_open_state(ops, sp);
		if (status < 0) {
			if (status == -NFS4ERR_NO_GRACE) {
				ops = &nfs4_network_partition_recovery_ops;
				status = nfs4_reclaim_open_state(ops, sp);
			}
			if (status == -NFS4ERR_STALE_CLIENTID)
				goto restart_loop;
			if (status == -NFS4ERR_EXPIRED)
				goto restart_loop;
		}
	}
	nfs_delegation_reap_unclaimed(clp);
out:
	up_write(&clp->cl_sem);
	unlock_kernel();
	if (status == -NFS4ERR_CB_PATH_DOWN)
		nfs_handle_cb_pathdown(clp);
T
Trond Myklebust 已提交
809
	nfs4_clear_recover_bit(clp);
810
	nfs_put_client(clp);
811
	module_put_and_exit(0);
L
Linus Torvalds 已提交
812 813 814
	return 0;
out_error:
	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %u.%u.%u.%u with error %d\n",
815
				NIPQUAD(clp->cl_addr.sin_addr), -status);
816
	set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
L
Linus Torvalds 已提交
817 818 819 820 821 822 823 824
	goto out;
}

/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */