nfs4state.c 24.7 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 45
/*
 *  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/config.h>
#include <linux/slab.h>
#include <linux/smp_lock.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_idmap.h>
46 47
#include <linux/kthread.h>
#include <linux/module.h>
L
Linus Torvalds 已提交
48 49 50
#include <linux/workqueue.h>
#include <linux/bitops.h>

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

#define OPENOWNER_POOL_SIZE	8

57
const nfs4_stateid zero_stateid;
L
Linus Torvalds 已提交
58

59
static DEFINE_SPINLOCK(state_spinlock);
L
Linus Torvalds 已提交
60 61 62 63 64 65 66 67 68 69 70 71
static LIST_HEAD(nfs4_clientid_list);

void
init_nfsv4_state(struct nfs_server *server)
{
	server->nfs4_state = NULL;
	INIT_LIST_HEAD(&server->nfs4_siblings);
}

void
destroy_nfsv4_state(struct nfs_server *server)
{
J
Jesper Juhl 已提交
72 73
	kfree(server->mnt_path);
	server->mnt_path = NULL;
L
Linus Torvalds 已提交
74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
	if (server->nfs4_state) {
		nfs4_put_client(server->nfs4_state);
		server->nfs4_state = NULL;
	}
}

/*
 * nfs4_get_client(): returns an empty client structure
 * nfs4_put_client(): drops reference to client structure
 *
 * Since these are allocated/deallocated very rarely, we don't
 * bother putting them in a slab cache...
 */
static struct nfs4_client *
nfs4_alloc_client(struct in_addr *addr)
{
	struct nfs4_client *clp;

	if (nfs_callback_up() < 0)
		return NULL;
94
	if ((clp = kzalloc(sizeof(*clp), GFP_KERNEL)) == NULL) {
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102 103 104 105 106 107
		nfs_callback_down();
		return NULL;
	}
	memcpy(&clp->cl_addr, addr, sizeof(clp->cl_addr));
	init_rwsem(&clp->cl_sem);
	INIT_LIST_HEAD(&clp->cl_delegations);
	INIT_LIST_HEAD(&clp->cl_state_owners);
	INIT_LIST_HEAD(&clp->cl_unused);
	spin_lock_init(&clp->cl_lock);
	atomic_set(&clp->cl_count, 1);
	INIT_WORK(&clp->cl_renewd, nfs4_renew_state, clp);
	INIT_LIST_HEAD(&clp->cl_superblocks);
	rpc_init_wait_queue(&clp->cl_rpcwaitq, "NFS4 client");
108
	clp->cl_rpcclient = ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
109
	clp->cl_boot_time = CURRENT_TIME;
110
	clp->cl_state = 1 << NFS4CLNT_LEASE_EXPIRED;
L
Linus Torvalds 已提交
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
	return clp;
}

static void
nfs4_free_client(struct nfs4_client *clp)
{
	struct nfs4_state_owner *sp;

	while (!list_empty(&clp->cl_unused)) {
		sp = list_entry(clp->cl_unused.next,
				struct nfs4_state_owner,
				so_list);
		list_del(&sp->so_list);
		kfree(sp);
	}
	BUG_ON(!list_empty(&clp->cl_state_owners));
	nfs_idmap_delete(clp);
128
	if (!IS_ERR(clp->cl_rpcclient))
L
Linus Torvalds 已提交
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
		rpc_shutdown_client(clp->cl_rpcclient);
	kfree(clp);
	nfs_callback_down();
}

static struct nfs4_client *__nfs4_find_client(struct in_addr *addr)
{
	struct nfs4_client *clp;
	list_for_each_entry(clp, &nfs4_clientid_list, cl_servers) {
		if (memcmp(&clp->cl_addr, addr, sizeof(clp->cl_addr)) == 0) {
			atomic_inc(&clp->cl_count);
			return clp;
		}
	}
	return NULL;
}

struct nfs4_client *nfs4_find_client(struct in_addr *addr)
{
	struct nfs4_client *clp;
	spin_lock(&state_spinlock);
	clp = __nfs4_find_client(addr);
	spin_unlock(&state_spinlock);
	return clp;
}

struct nfs4_client *
nfs4_get_client(struct in_addr *addr)
{
	struct nfs4_client *clp, *new = NULL;

	spin_lock(&state_spinlock);
	for (;;) {
		clp = __nfs4_find_client(addr);
		if (clp != NULL)
			break;
		clp = new;
		if (clp != NULL) {
			list_add(&clp->cl_servers, &nfs4_clientid_list);
			new = NULL;
			break;
		}
		spin_unlock(&state_spinlock);
		new = nfs4_alloc_client(addr);
		spin_lock(&state_spinlock);
		if (new == NULL)
			break;
	}
	spin_unlock(&state_spinlock);
	if (new)
		nfs4_free_client(new);
	return clp;
}

void
nfs4_put_client(struct nfs4_client *clp)
{
	if (!atomic_dec_and_lock(&clp->cl_count, &state_spinlock))
		return;
	list_del(&clp->cl_servers);
	spin_unlock(&state_spinlock);
	BUG_ON(!list_empty(&clp->cl_superblocks));
	rpc_wake_up(&clp->cl_rpcwaitq);
	nfs4_kill_renewd(clp);
	nfs4_free_client(clp);
}

196
static int nfs4_init_client(struct nfs4_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
197
{
198 199
	int status = nfs4_proc_setclientid(clp, NFS4_CALLBACK,
			nfs_callback_tcpport, cred);
L
Linus Torvalds 已提交
200
	if (status == 0)
201
		status = nfs4_proc_setclientid_confirm(clp, cred);
L
Linus Torvalds 已提交
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
	if (status == 0)
		nfs4_schedule_state_renewal(clp);
	return status;
}

u32
nfs4_alloc_lockowner_id(struct nfs4_client *clp)
{
	return clp->cl_lockowner_id ++;
}

static struct nfs4_state_owner *
nfs4_client_grab_unused(struct nfs4_client *clp, struct rpc_cred *cred)
{
	struct nfs4_state_owner *sp = NULL;

	if (!list_empty(&clp->cl_unused)) {
		sp = list_entry(clp->cl_unused.next, struct nfs4_state_owner, so_list);
		atomic_inc(&sp->so_count);
		sp->so_cred = cred;
		list_move(&sp->so_list, &clp->cl_state_owners);
		clp->cl_nunused--;
	}
	return sp;
}

228 229 230 231 232 233 234 235 236 237 238 239 240 241
struct rpc_cred *nfs4_get_renew_cred(struct nfs4_client *clp)
{
	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;
}

242 243 244 245 246 247 248 249 250 251 252 253
struct rpc_cred *nfs4_get_setclientid_cred(struct nfs4_client *clp)
{
	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
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
static struct nfs4_state_owner *
nfs4_find_state_owner(struct nfs4_client *clp, struct rpc_cred *cred)
{
	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;

281
	sp = kzalloc(sizeof(*sp),GFP_KERNEL);
L
Linus Torvalds 已提交
282 283
	if (!sp)
		return NULL;
284
	spin_lock_init(&sp->so_lock);
L
Linus Torvalds 已提交
285 286
	INIT_LIST_HEAD(&sp->so_states);
	INIT_LIST_HEAD(&sp->so_delegations);
287 288 289 290
	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
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
	atomic_set(&sp->so_count, 1);
	return sp;
}

void
nfs4_drop_state_owner(struct nfs4_state_owner *sp)
{
	struct nfs4_client *clp = sp->so_client;
	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)
{
	struct nfs4_client *clp = server->nfs4_state;
	struct nfs4_state_owner *sp, *new;

	get_rpccred(cred);
	new = nfs4_alloc_state_owner();
	spin_lock(&clp->cl_lock);
	sp = nfs4_find_state_owner(clp, cred);
	if (sp == NULL)
		sp = nfs4_client_grab_unused(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);
		new->so_cred = cred;
		sp = new;
		new = NULL;
	}
	spin_unlock(&clp->cl_lock);
J
Jesper Juhl 已提交
328
	kfree(new);
L
Linus Torvalds 已提交
329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
	if (sp != NULL)
		return sp;
	put_rpccred(cred);
	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)
{
	struct nfs4_client *clp = sp->so_client;
	struct rpc_cred *cred = sp->so_cred;

	if (!atomic_dec_and_lock(&sp->so_count, &clp->cl_lock))
		return;
	if (clp->cl_nunused >= OPENOWNER_POOL_SIZE)
		goto out_free;
	if (list_empty(&sp->so_list))
		goto out_free;
	list_move(&sp->so_list, &clp->cl_unused);
	clp->cl_nunused++;
	spin_unlock(&clp->cl_lock);
	put_rpccred(cred);
	cred = NULL;
	return;
out_free:
	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;

368
	state = kzalloc(sizeof(*state), GFP_KERNEL);
L
Linus Torvalds 已提交
369 370 371 372
	if (!state)
		return NULL;
	atomic_set(&state->count, 1);
	INIT_LIST_HEAD(&state->lock_states);
373
	spin_lock_init(&state->state_lock);
L
Linus Torvalds 已提交
374 375 376
	return state;
}

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
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
394 395 396 397 398 399 400 401
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? */
402
		if (state->state == 0)
L
Linus Torvalds 已提交
403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429
			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();
430
	spin_lock(&owner->so_lock);
L
Linus Torvalds 已提交
431 432 433 434 435 436 437 438 439
	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);
440 441 442 443
		/* 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
444 445
	} else {
		spin_unlock(&inode->i_lock);
446
		spin_unlock(&owner->so_lock);
L
Linus Torvalds 已提交
447 448 449 450 451 452 453 454 455
		if (new)
			nfs4_free_open_state(new);
	}
out:
	return state;
}

/*
 * Beware! Caller must be holding exactly one
456
 * reference to clp->cl_sem!
L
Linus Torvalds 已提交
457 458 459 460 461 462
 */
void nfs4_put_open_state(struct nfs4_state *state)
{
	struct inode *inode = state->inode;
	struct nfs4_state_owner *owner = state->owner;

463
	if (!atomic_dec_and_lock(&state->count, &owner->so_lock))
L
Linus Torvalds 已提交
464
		return;
465
	spin_lock(&inode->i_lock);
L
Linus Torvalds 已提交
466 467 468
	if (!list_empty(&state->inode_states))
		list_del(&state->inode_states);
	list_del(&state->open_states);
469 470
	spin_unlock(&inode->i_lock);
	spin_unlock(&owner->so_lock);
L
Linus Torvalds 已提交
471 472 473 474 475 476
	iput(inode);
	nfs4_free_open_state(state);
	nfs4_put_state_owner(owner);
}

/*
477
 * Close the current file.
L
Linus Torvalds 已提交
478 479 480 481 482
 */
void nfs4_close_state(struct nfs4_state *state, mode_t mode)
{
	struct inode *inode = state->inode;
	struct nfs4_state_owner *owner = state->owner;
483
	int oldstate, newstate = 0;
L
Linus Torvalds 已提交
484 485 486

	atomic_inc(&owner->so_count);
	/* Protect against nfs4_find_state() */
487
	spin_lock(&owner->so_lock);
L
Linus Torvalds 已提交
488
	spin_lock(&inode->i_lock);
489 490 491 492 493 494 495 496 497 498
	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--;
	}
499
	oldstate = newstate = state->state;
500 501 502 503 504 505
	if (state->n_rdwr == 0) {
		if (state->n_rdonly == 0)
			newstate &= ~FMODE_READ;
		if (state->n_wronly == 0)
			newstate &= ~FMODE_WRITE;
	}
506 507 508
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		nfs4_state_set_mode_locked(state, newstate);
		oldstate = newstate;
L
Linus Torvalds 已提交
509 510
	}
	spin_unlock(&inode->i_lock);
511
	spin_unlock(&owner->so_lock);
512 513 514

	if (oldstate != newstate && nfs4_do_close(inode, state) == 0)
		return;
L
Linus Torvalds 已提交
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
	nfs4_put_open_state(state);
	nfs4_put_state_owner(owner);
}

/*
 * 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;
	struct nfs4_client *clp = state->owner->so_client;

546
	lsp = kzalloc(sizeof(*lsp), GFP_KERNEL);
L
Linus Torvalds 已提交
547 548
	if (lsp == NULL)
		return NULL;
549
	lsp->ls_seqid.sequence = &state->owner->so_sequence;
L
Linus Torvalds 已提交
550 551 552 553 554
	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);
555
	INIT_LIST_HEAD(&lsp->ls_locks);
L
Linus Torvalds 已提交
556 557 558 559 560 561 562
	return lsp;
}

/*
 * Return a compatible lock_state. If no initialized lock_state structure
 * exists, return an uninitialized one.
 *
563
 * The caller must be holding clp->cl_sem
L
Linus Torvalds 已提交
564
 */
565
static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
L
Linus Torvalds 已提交
566
{
567
	struct nfs4_lock_state *lsp, *new = NULL;
L
Linus Torvalds 已提交
568
	
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
	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
589 590 591 592
	return lsp;
}

/*
593 594
 * Release reference to lock_state, and free it if we see that
 * it is no longer in use
L
Linus Torvalds 已提交
595
 */
596
void nfs4_put_lock_state(struct nfs4_lock_state *lsp)
L
Linus Torvalds 已提交
597
{
598
	struct nfs4_state *state;
L
Linus Torvalds 已提交
599

600 601 602 603 604 605 606 607 608 609
	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
610 611
}

612
static void nfs4_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
L
Linus Torvalds 已提交
613
{
614
	struct nfs4_lock_state *lsp = src->fl_u.nfs4_fl.owner;
L
Linus Torvalds 已提交
615

616 617 618
	dst->fl_u.nfs4_fl.owner = lsp;
	atomic_inc(&lsp->ls_count);
}
L
Linus Torvalds 已提交
619

620
static void nfs4_fl_release_lock(struct file_lock *fl)
L
Linus Torvalds 已提交
621
{
622
	nfs4_put_lock_state(fl->fl_u.nfs4_fl.owner);
L
Linus Torvalds 已提交
623 624
}

625 626 627 628 629 630
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
631
{
632 633 634 635 636 637 638 639 640 641
	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
642 643
}

644 645 646
/*
 * Byte-range lock aware utility to initialize the stateid of read/write
 * requests.
L
Linus Torvalds 已提交
647
 */
648
void nfs4_copy_stateid(nfs4_stateid *dst, struct nfs4_state *state, fl_owner_t fl_owner)
L
Linus Torvalds 已提交
649
{
650
	struct nfs4_lock_state *lsp;
L
Linus Torvalds 已提交
651

652 653 654
	memcpy(dst, &state->stateid, sizeof(*dst));
	if (test_bit(LK_STATE_IN_USE, &state->flags) == 0)
		return;
L
Linus Torvalds 已提交
655

656 657 658 659 660
	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
661 662 663
	nfs4_put_lock_state(lsp);
}

664 665
struct nfs_seqid *nfs_alloc_seqid(struct nfs_seqid_counter *counter)
{
666
	struct rpc_sequence *sequence = counter->sequence;
667 668 669 670 671
	struct nfs_seqid *new;

	new = kmalloc(sizeof(*new), GFP_KERNEL);
	if (new != NULL) {
		new->sequence = counter;
672 673 674
		spin_lock(&sequence->lock);
		list_add_tail(&new->list, &sequence->list);
		spin_unlock(&sequence->lock);
675 676 677 678 679
	}
	return new;
}

void nfs_free_seqid(struct nfs_seqid *seqid)
L
Linus Torvalds 已提交
680
{
681 682
	struct rpc_sequence *sequence = seqid->sequence->sequence;

683 684 685 686
	spin_lock(&sequence->lock);
	list_del(&seqid->list);
	spin_unlock(&sequence->lock);
	rpc_wake_up(&sequence->wait);
687
	kfree(seqid);
L
Linus Torvalds 已提交
688 689 690
}

/*
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
 * 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()
 */
static inline void nfs_increment_seqid(int status, struct nfs_seqid *seqid)
{
	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
722
		nfs4_drop_state_owner(sp);
723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
	}
	return nfs_increment_seqid(status, seqid);
}

/*
 * 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)
{
	return nfs_increment_seqid(status, seqid);
}

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

742 743
	if (sequence->list.next == &seqid->list)
		goto out;
744
	spin_lock(&sequence->lock);
745
	if (sequence->list.next != &seqid->list) {
746 747
		rpc_sleep_on(&sequence->wait, task, NULL, NULL);
		status = -EAGAIN;
748
	}
749
	spin_unlock(&sequence->lock);
750
out:
751
	return status;
L
Linus Torvalds 已提交
752 753 754 755
}

static int reclaimer(void *);

T
Trond Myklebust 已提交
756 757 758 759 760 761 762 763 764
static inline void nfs4_clear_recover_bit(struct nfs4_client *clp)
{
	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
765 766 767
/*
 * State recovery routine
 */
768
static void nfs4_recover_state(struct nfs4_client *clp)
L
Linus Torvalds 已提交
769
{
770
	struct task_struct *task;
L
Linus Torvalds 已提交
771

772 773 774 775 776 777
	__module_get(THIS_MODULE);
	atomic_inc(&clp->cl_count);
	task = kthread_run(reclaimer, clp, "%u.%u.%u.%u-reclaim",
			NIPQUAD(clp->cl_addr));
	if (!IS_ERR(task))
		return;
T
Trond Myklebust 已提交
778
	nfs4_clear_recover_bit(clp);
779 780
	nfs4_put_client(clp);
	module_put(THIS_MODULE);
L
Linus Torvalds 已提交
781 782 783 784 785
}

/*
 * Schedule a state recovery attempt
 */
786
void nfs4_schedule_state_recovery(struct nfs4_client *clp)
L
Linus Torvalds 已提交
787 788 789
{
	if (!clp)
		return;
T
Trond Myklebust 已提交
790
	if (test_and_set_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
791
		nfs4_recover_state(clp);
L
Linus Torvalds 已提交
792 793 794 795 796 797 798 799 800
}

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) {
801
		if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
L
Linus Torvalds 已提交
802 803 804 805 806 807 808 809 810 811 812 813 814 815
			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:
816
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
L
Linus Torvalds 已提交
817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882
				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;
}

883 884 885 886 887 888 889 890 891 892
static void nfs4_state_mark_reclaim(struct nfs4_client *clp)
{
	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;
893
		spin_lock(&sp->so_lock);
894 895 896 897 898 899 900
		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;
			}
		}
901
		spin_unlock(&sp->so_lock);
902 903 904
	}
}

L
Linus Torvalds 已提交
905 906
static int reclaimer(void *ptr)
{
907
	struct nfs4_client *clp = ptr;
L
Linus Torvalds 已提交
908 909
	struct nfs4_state_owner *sp;
	struct nfs4_state_recovery_ops *ops;
910
	struct rpc_cred *cred;
L
Linus Torvalds 已提交
911 912 913 914 915 916 917 918 919 920 921
	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:
922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942
	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 */
943
	nfs4_state_mark_reclaim(clp);
944 945 946 947 948
	status = -ENOENT;
	if (cred != NULL) {
		status = nfs4_init_client(clp, cred);
		put_rpccred(cred);
	}
L
Linus Torvalds 已提交
949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972
	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
973
	nfs4_clear_recover_bit(clp);
L
Linus Torvalds 已提交
974
	nfs4_put_client(clp);
975
	module_put_and_exit(0);
L
Linus Torvalds 已提交
976 977 978 979 980 981 982 983 984 985 986 987
	return 0;
out_error:
	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %u.%u.%u.%u with error %d\n",
				NIPQUAD(clp->cl_addr.s_addr), -status);
	goto out;
}

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